Journals >Laser & Optoelectronics Progress
Contents
2025
Volume: 62 Issue 8
43 Article(s)
Export citation format
Digital Image Processing
HSV Space-Based Nonlinear Adaptive Low-Light Image Enhancement Algorithm
Chengkang Yu, and Guangliang Han
To address issues such as low brightness, low contrast, and lack of details in images taken under challenging conditions like nighttime, backlighting, and severe weather, a nonlinear adaptive dark detail enhancement algorithm is proposed for improving low-light images. To ensure color authenticity, the original image iTo address issues such as low brightness, low contrast, and lack of details in images taken under challenging conditions like nighttime, backlighting, and severe weather, a nonlinear adaptive dark detail enhancement algorithm is proposed for improving low-light images. To ensure color authenticity, the original image is first converted to HSV space, and the brightness component V is extracted. For dealing with the issues of poor brightness and low contrast in low-light images, an improved gamma correction algorithm is then adopted to adaptively adjust image brightness. Subsequently, a brightness adaptive contrast enhancement algorithm is introduced, combining a low-pass filtering approach to adaptively enhance high-frequency details. This helps highlight textures and edge information of dark areas of the image. Finally, a brightness-guided adaptive image fusion algorithm is proposed to preserve edge details in highlighted areas while avoiding overexposure. Experimental results demonstrate that the proposed algorithm effectively adapts to the image characteristics of low-light environments. It not only significantly enhances the brightness and contrast of low-light images but also highlights details in darker areas while preserving color authenticity..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0837001 (2025)
Three-Dimensional Facial UV Texture Restoration Based on Gated Convolution
Zhenhua Zhao, Bo Yang, Qiuhang Chen, and Yiming Ying
In this study, a three-dimensional (3D) facial UV texture restoration algorithm based on gated convolution is proposed to address the texture loss caused by self-occlusion in unconstrained facial images captured from large viewing angles during reconstructing 3D facial structures from a single image. First, a gated conIn this study, a three-dimensional (3D) facial UV texture restoration algorithm based on gated convolution is proposed to address the texture loss caused by self-occlusion in unconstrained facial images captured from large viewing angles during reconstructing 3D facial structures from a single image. First, a gated convolution mechanism is designed to learn a dynamic feature selection approach for each channel and spatial position, thereby enhancing the network's ability to capture complex nonlinear features. These gated convolutions are then stacked to form an encoder-decoder network that repairs 3D facial UV texture images with irregular defects. In addition, a spectral normalization loss function is introduced to stabilize the generative adversarial network, and a segmented training approach is implemented to overcome the challenges of cost and accessibility in collecting 3D facial texture datasets. The experimental results show that the proposed algorithm outperforms mainstream algorithms in terms of the peak signal-to-noise ratio and structural similarity. The proposed algorithm effectively restores UV texture maps under large angle occlusion, yielding more comprehensive facial texture maps with natural, coherent pixel restoration, and realistic texture details..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0837002 (2025)
Cotter Pin Detection of Transmission Line Based on YOLOv8-DEA
Kerui Wang, Chuncheng Zhou, Yi Ma, Lincong Peng, Hao Zhou, and Pengfei Yu
Due to the complex background and small target of the transmission line cotter pin, the intelligent power inspection of unmanned aerial vehicles (UAVs) is vulnerable to the problems of low detection accuracy and high missed and false detection rates. Addressing these issues, the present study proposes a target detectioDue to the complex background and small target of the transmission line cotter pin, the intelligent power inspection of unmanned aerial vehicles (UAVs) is vulnerable to the problems of low detection accuracy and high missed and false detection rates. Addressing these issues, the present study proposes a target detection algorithm based on YOLOv8-DEA to better adapt to UAVs and other application scenarios. First, the backbone network's C2f module is modified, enabling the model to focus on regions of interest and enhancing its perception of local image structures. Subsequently, an efficient mamba-like linear attention (EMLLA) mechanism is used to capture distant dependencies, and the efficient multilayer perceptron (EMLP) module is applied to map the model features to a higher dimensionality, enhancing the model's expressiveness. Finally, a dynamic selection mechanism is used to improve the Neck layer. The adaptive fusion of deep and shallow features enables the effective integration of features from different levels, allowing the model to accurately capture global semantic information, as well as extract rich detailed information when processing complex and diverse data. The experimental results demonstrate that the improved algorithm achieves a 2.33 percentage points increase in mean average precision (mAP@0.5) and 3.67 percentage points increase in recall (R@0.5) on the custom cotter pin dataset. Additionally, the algorithm achieves a precision of 95.58% and speed of 67.84 frame/s. When compared to mainstream algorithms, the proposed method not only exhibits improved detection accuracy, but also ensures real-time performance, making it more suitable for the needs of transmission-line cotter pin detection in engineering applications..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0837003 (2025)
Image Dehazing Algorithm Based on Multi-Dimensional Attention Feature Fusion
Xuguang Zhu, and Nan Jiang
To address the problems of detail loss and color distortion in the current image defogging algorithms, this paper proposes a multi-dimensional attention feature fusion image dehazing algorithm. The core step of the proposed algorithm is the introduction of a union attention mechanism module, which can simultaneously opTo address the problems of detail loss and color distortion in the current image defogging algorithms, this paper proposes a multi-dimensional attention feature fusion image dehazing algorithm. The core step of the proposed algorithm is the introduction of a union attention mechanism module, which can simultaneously operate in three dimensions of channel, space, and pixel to achieve accurate enhancement of local features, while parallel a multi-scale perceptual feature fusion module effectively captures global feature information of different scales. To achieve a more refined and accurate dehazing effect, a bi-directional gated feature fusion mechanism is added to the proposed algorithm to realize the deep fusion and complementarity of local information and global information features. Experimental validation on multiple datasets, such as RESIDE, I-Hazy, and O-Hazy shows that, the proposed algorithm exhibits better performance than the existing state-of-the-art in terms of peak signal-to-noise ratio (PSNR) and structural similarity (SSIM). Compared with the classical GCA-Net, the PSNR and SSIM of the proposed algorithm increased by 2.77 dB and 0.0046, respectively. Results of this study can provide new insights and directions for investigating image dehazing algorithms..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0837004 (2025)
Research on Floc Feature Detection Method Based on Improved Density Map and Local Enhanced CNN
Jie Luo, and Junran Zhang
In the context of real-world observations of water purification flocculation processes, current image segmentation-based methods for detecting floc features face several challenges, which include poor recognition accuracy for deep-lying flocs, high annotation costs, and difficulties in adaptively processing depth-of-fiIn the context of real-world observations of water purification flocculation processes, current image segmentation-based methods for detecting floc features face several challenges, which include poor recognition accuracy for deep-lying flocs, high annotation costs, and difficulties in adaptively processing depth-of-field information aiming at these problems, a new floc feature detection method based on improved density map and locally enhanced convolutional neural network (LECNN) is proposed. First, a density map construction method based on multipoint marking and average kernel smoothing is designed, to address the inability of the density map to simultaneously reflect multiple floc feature parameters. Second, a scene depth adaptive structure that assigns different weights to flocs at various depths is proposed, to mitigate the inaccuracies in floc parameter detection caused by parallax. Then, the proposed LECNN captures multiscale receptive fields while emphasizing local features. In comparative tests on a floc image dataset with multipoint markings, LECNN demonstrates accurate and robust density map fitting performance against recently proposed pixel-level prediction network structures, achieving a performance improvement over other floc feature detection benchmark methods in experimental results..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0837005 (2025)
Weakly-Supervised Point Cloud Semantic Segmentation with Consistency Constraint and Feature Enhancement
Dong Wei, Yifan Bai, He Sun, and Jingtian Zhang
To address issues such as poor output consistency, information loss, and blurred boundaries caused by incomplete truth labeling in current weakly-supervised point cloud semantic segmentation methods, a weakly-supervised point cloud semantic segmentation method with input consistency constraint and feature enhancement iTo address issues such as poor output consistency, information loss, and blurred boundaries caused by incomplete truth labeling in current weakly-supervised point cloud semantic segmentation methods, a weakly-supervised point cloud semantic segmentation method with input consistency constraint and feature enhancement is proposed. Additional constraint is provided on the input point cloud to learn the input consistency of the augmented point cloud data, in order to better understand the essential features of the data and improve the generalization ability of the model. An adaptive enhancement mechanism is introduced in the point feature extractor to enhance the model's perceptual ability, and utilizing sub scene boundary contrastive optimization to further improve the segmentation accuracy of the boundary region. By utilizing query operations in point feature query network, sparse training signals are fully utilized, and a channel attention mechanism module is constructed to enhance the representation ability of important features by strengthening channel dependencies, resulting in more effective prediction of point cloud semantic labels. Experimental results show that the proposed method achieves good segmentation performance on three public point cloud datasets of S3DIS, Semantic3D, and Toronto3D, with a mean intersection over union of 66.4%, 77.9%, and 80.5%, respectively, using 1.0% truth labels for training..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0837006 (2025)
Study on the Decision Tree Model for Carbonate Rock Lithology Identification Based on Hyperspectral Data
Yu Huang, Yanlin Shao, Wei Wei, and Qihong Zeng
Hyperspectral remote sensing has been widely applied in geological research due to its rich multi-band spectral information. Most studies mainly focus on the identification of soil components and clay minerals, with relatively fewer studies on carbonate rocks, so this paper proposes a decision tree model to achieve preHyperspectral remote sensing has been widely applied in geological research due to its rich multi-band spectral information. Most studies mainly focus on the identification of soil components and clay minerals, with relatively fewer studies on carbonate rocks, so this paper proposes a decision tree model to achieve precise classification of carbonate rocks based on hyperspectral data. A continuum-removed method is used to preprocess the data, and then combines spectral knowledge and machine learning to extract features. Specifically, the study determines spectral intervals closely related to carbonate rocks through spectral knowledge and extracts key waveform features from the spectral curves. Subsequently, the study uses the random forest algorithm to select features with discriminative capabilities, determines the optimal classification discriminant through threshold analysis, and builds a decision tree model. Finally, the model performance is evaluated using a confusion matrix, and the classification accuracy is compared with other five models. Results show that the decision tree model constructed based on the order of the lowest point wavelength of the absorption valley, the right shoulder wavelength of the absorption band , and the absorption bandwidth exhibited the highest classification accuracy, with an accuracy rate of 95.57%..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0837007 (2025)
Window-Detection Method Based on Hole Constraints and Hierarchical Localization
Xiaojuan Ning, Jiawei Du, Chunxu Li, Lei Huang, Zhenghao Shi, and Haiyan Jin
To address the issue of low window-detection completeness and accuracy caused by the irregular distribution of windows on building facades, this study proposes a novel window-detection method that leverages hole constraints and hierarchical localization. This approach utilizes the least-squares method to fit lines to tTo address the issue of low window-detection completeness and accuracy caused by the irregular distribution of windows on building facades, this study proposes a novel window-detection method that leverages hole constraints and hierarchical localization. This approach utilizes the least-squares method to fit lines to the projected point cloud data of the building facade, with distance constraints applied to obtain the primary wall point cloud data. The initial window position is determined using the hole-based detection method. Incorporating the concept of region expansion, the method employs an improved Alpha-Shape algorithm to extract boundary points around the initially identified window positions. Feature points among the boundary points are identified, and the boundary points are regularized based on these feature points, thus enabling the precise construction of window wireframe models. Experimental results show that this method significantly improves the accuracy of window detection, as evidenced by its average accuracy and completeness of 100% and 93.34%, respectively..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0837008 (2025)
Improved 3D Reconstruction Algorithm for Unmanned Aerial Vehicle Images Based on PM-MVS
Peixin He, Xiyan Sun, Yuanfa Ji, Yang Bai, and Yu Chen
To address the challenges of long reconstruction time and numerous model voids in large-scale scenes and weakly textured regions during 3D reconstruction of unmanned aerial vehicle (UAV) images using existing multi-view stereo reconstruction (MVS) algorithms, an improved 3D reconstruction algorithm based on PatchMatch To address the challenges of long reconstruction time and numerous model voids in large-scale scenes and weakly textured regions during 3D reconstruction of unmanned aerial vehicle (UAV) images using existing multi-view stereo reconstruction (MVS) algorithms, an improved 3D reconstruction algorithm based on PatchMatch MVS (PM-MVS), called MCP-MVS, is proposed. The algorithm employs a multi-constraint matching cost computation method to eliminate outlier points from the 3D point cloud, thereby enhancing robustness. A pyramid red-and-black checkerboard sampling propagation strategy is introduced to extract geometric features across different scale spaces, while graphics processing unit based parallel propagation is exploited to improve the reconstruction efficiency. Experiments conducted on three UAV datasets demonstrate that MCP-MVS improves reconstruction efficiency by at least 16.6% compared to state-of-the-art algorithms, including PMVS, Colmap, OpenMVS, and 3DGS. Moreover, on the Cadastre dataset, the overall error is reduced by 35.7%, 20.3%, 19.5%, and 11.6% compared to PMVS, Colmap, OpenMVS, and 3DGS, respectively. The proposed algorithm also achieves the highest F-scores on the Cadastre and GDS datasets, 75.76% and 79.02%, respectively. These results demonstrate that the proposed algorithm significantly reduces model voids, validating its effectiveness and practicality..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0837009 (2025)
Imaging Systems
Focal Stack-Based Light Field Reconstruction and Measurement
Zhiting Zhang, Yang Liu, Sibo Huang, Wanyu Gu, Xiaoli Liu, Xiang Peng, and Zewei Cai
The focal stack-based light field reconstruction method uses a camera to capture an image stack at various focal planes in image space, thereby reconstructing a full-resolution light field based on the transport-of-intensity property of the light field and transmission distance in object space. However, the nonlinear dThe focal stack-based light field reconstruction method uses a camera to capture an image stack at various focal planes in image space, thereby reconstructing a full-resolution light field based on the transport-of-intensity property of the light field and transmission distance in object space. However, the nonlinear depth mapping between object and image spaces combines with image distortion introduce differences between the image acquisition in image space and light field reconstruction in object space, leading to decreased accuracy in reconstruction. This study first analyzes the consistency of focal stack-based light field reconstruction. A pre-calibrated light field camera is used to establish a metric mapping relationship between object and image spaces and the reference light field data. The multiview imaging and depth range of the reconstructed light field are then quantitatively analyzed. Subsequently, a consistent focal stack-based light field reconstruction method is proposed by establishing an accurate object-image space conversion relationship in the focal scanning space through camera calibration. The image stack in image space can be converted into object space, to uniformly eliminate image distortion for consistent light field reconstruction. Finally, depth measurement in the focal scanning space is achieved through the reconstructed light fields in object space..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0811001 (2025)
Dynamic Laser Illumination for Silicon-Based Liquid-Crystal Projection System
Hongxin Li, Xianpeng Zhang, and Yatao Yang
A projection display module with compact size, low power consumption, and high resolution is key for augmented-reality glasses. To reduce the power consumption of the projection system, a silicon-based liquid-crystal projection system with dynamic illumination based on laser-beam scanning is designed. The system adoptsA projection display module with compact size, low power consumption, and high resolution is key for augmented-reality glasses. To reduce the power consumption of the projection system, a silicon-based liquid-crystal projection system with dynamic illumination based on laser-beam scanning is designed. The system adopts a front-illumination architecture to realize spatial conversion between the laser-beam scanning angle and the imaging position of the spatial light modulator by multiplexing lenses. Additionally, it realizes 638 partitions in a 30°×40° field of view. By modeling the laser illumination intensity of individual partitions and simulating the dynamic illumination light-field distribution under different display scenarios, a partitioned illumination model of laser-beam scanning is established. Simulation results show that the dynamic illumination of laser-beam scanning can reduce lighting power consumption to less than 10% of power consumed by a non-dynamic illumination system in typical application scenarios of augmented reality, such as information prompting and navigation..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0811002 (2025)
Design of Image-Acquisition System for I-ToF Image Sensor Based on FPGA
Qian Wan, Jiangtao Xu, Quanmin Chen, Zijian Liu, and Yijie Zhang
As conventional image-acquisition systems cannot easily fulfill the demand for rapid transmission of significant data amounts from indirect time-of-flight (I-ToF) image sensors, a high-speed image-acquisition system based on a field-programmable gate array (FPGA) and multilane camera serial interface 2 (CSI-2) is propoAs conventional image-acquisition systems cannot easily fulfill the demand for rapid transmission of significant data amounts from indirect time-of-flight (I-ToF) image sensors, a high-speed image-acquisition system based on a field-programmable gate array (FPGA) and multilane camera serial interface 2 (CSI-2) is proposed. Based on the CSI-2 protocol, the system adopts a digital physical layer with 12 independent lanes, which can significantly improve the acquisition rate of sensor data. Simultaneously, the system utilizes the short-packet information of the CSI-2 protocol to denote different phase-shift images output by the I-ToF sensor, thus enabling the host computer to reconstruct depth images accurately. In the FPGA design, the low level protocol layer module is designed based on a parallel architecture to process CSI-2 data packets at high speed. Additionally, a parallel first-in first-out (FIFO) structure is adopted to eliminate delays between different data lanes. Experimental results show that the system can support the complete acquisition of I-ToF image sensor data with a 2016 pixel×1096 pixel resolution and 30 frame/s frame rate. In the high-speed burst mode of the CSI-2, the instantaneous bandwidth of the FPGA data acquisition can reach a maximum of 9.600 Gb/s. The depth image reconstructed by the host computer enables distance measurement. Therefore, the image-acquisition system based on FPGA and multilane CSI-2 enables high-speed I-ToF image-sensor data transmission..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0811003 (2025)
Surface Structured Light Decoding Algorithm Based on Illumination Intensity
Jiaxuan Han, Yi Qian, Yang Yang, and Shigang Liu
To address the problem wherein the existing decode methods do not consider the effect of global illumination on decoding accuracy, this paper proposes a surface structured light decoding algorithm based on illumination intensity. First, the appropriate subhigh-frequency patterns from a series of encoded patterns are seTo address the problem wherein the existing decode methods do not consider the effect of global illumination on decoding accuracy, this paper proposes a surface structured light decoding algorithm based on illumination intensity. First, the appropriate subhigh-frequency patterns from a series of encoded patterns are selected to estimate the ratio coefficient of the direct and global illumination intensities for a scene. Subsequently, the ratio coefficient is used to compute the direct and global light intensities of each pixel point. Finally, a decoding rule is constructed based on the direct and global light intensities, followed by three-dimensional reconstruction. Comparative experimental results show that the proposed decoding algorithm can effectively improve both the decoding accuracy and quality of the reconstructed point cloud..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0811004 (2025)
Lightweight Unsupervised Monocular Depth Estimation Framework Using Attention Mechanisms
Xiyu Li, Yilihamu Yaermaimaiti, Lirong Xie, and Shuoqi Cheng
To overcome the inherent limitations of existing unsupervised monocular depth estimation frameworks and enhance the network's generalizability across various scenarios, a lightweight unsupervised monocular depth estimation method that combines convolutional neural networks, attention mechanisms, and speeded-up robuTo overcome the inherent limitations of existing unsupervised monocular depth estimation frameworks and enhance the network's generalizability across various scenarios, a lightweight unsupervised monocular depth estimation method that combines convolutional neural networks, attention mechanisms, and speeded-up robust features (SURF) is proposed. First, a residual block with a linear self-attention mechanism (CCT-Block) and a residual block with a coordinate attention mechanism (CA-Block) were designed. These residual blocks were alternately used within the residual network framework to construct a multiscale encoder capable of capturing rich contextual information and mapping the relationship between the depth and image features while reducing the requirements for parameter computation and storage. In addition, the reprojection error of SURF was introduced to mitigate ambiguities that may arise in depth and pose estimation networks. Finally, evaluations were conducted on multiple datasets, including KITTI, Make3D, NYUDepth-v2, and Cityscapes. The experimental results show that the proposed method achieves an absolute relative error of 0.107 and a root mean square error of 4.674 on the KITTI dataset using only 4.9×106 model parameters. Furthermore, the proposed method exhibits strong generalizability across different datasets..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0811005 (2025)
Wider Field of View Imaging Method in Super-Resolution Microscopy Based on Square Aperture
Bowen Liu, Junkang Dai, Zhen'an Fu, Zitong Jin, Minghui Duan, Huaian Chen, and Yi Jin
In the field of microscopy, achieving a wider field of view and higher resolution are two critical goals pursued by researchers. Structured illumination microscopy (SIM) has effectively addressed the demand for higher resolution by surpassing the optical diffraction limit. However, limitations in beam range within struIn the field of microscopy, achieving a wider field of view and higher resolution are two critical goals pursued by researchers. Structured illumination microscopy (SIM) has effectively addressed the demand for higher resolution by surpassing the optical diffraction limit. However, limitations in beam range within structured illumination systems often cause pixel aliasing image artifacts when increasing the field of view. To eliminate these artifacts, we combine the coded aperture imaging technique, commonly used in computational imaging, with structured illumination microscopy, thereby proposing a super-resolution microscopy method that utilizes a square aperture to enhance the field of view. The square aperture is used to recover mixed frequency domain information in images captured with a 60× objective lens, ultimately reconstructing images that achieve the same resolving power as a 100× objective lens while attaining a fourfold increase in the field of view. The effectiveness of this method is demonstrated through both simulated and actual imaging data. This study provides a novel approach for enhancing the field of view in super-resolution structured illumination microscopy systems..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0811006 (2025)
Instrumentation, Measurement and Metrology
Failure Detection Algorithm for Electric Multiple Unit Brake Disc Bolts Based on SSD-YOLO
Yehong Chen, Hang Zhou, Xin Lu, Jia Yu, Ruiyu Han, Yifan Zhang, Tu Lü, Hui Wang, Tengyu Zhang, Yi Liu, and Kewei Song
Manual failure detection in the train of electric multiple unit (EMU) failure detection system (TEDS) involves high error rates and significant workload. To address this problem, this paper proposes an SSD-YOLO algorithm for detection of EMU brake disc bolts failures, which is improved from YOLOv5n. The network structuManual failure detection in the train of electric multiple unit (EMU) failure detection system (TEDS) involves high error rates and significant workload. To address this problem, this paper proposes an SSD-YOLO algorithm for detection of EMU brake disc bolts failures, which is improved from YOLOv5n. The network structure of YOLOv5n is improved considering that the failure areas of brake disc bolts are small and the failures are similar to that of normal samples. To make the model flexibly adapt to the features of different scales and enlarge the receptive field, the down sampling convolution layers in the backbone network are replaced with switchable atrous convolution layers. To enhance the ability of the network to obtain global information and interact with contextual information, this study integrates the Swin-Transformer module at the end of the backbone network. To better deal with semantic information of different scales and resolutions, the coupled detection head of the YOLO series is replaced with an efficient decoupled head, which can extract target location and category information. To further improve the training convergence speed of the algorithm, the SCYLLA-intersection over union (SIoU) loss function, which has better positioning ability, is used to replace the CIoU loss function. In this study, artificially annotated samples of the EMU brake disc bolts are used to train the network. Experiments show that the detection mean average precision (mAP) value of the improved algorithm on the EMU brake disc bolts failure dataset increased by 6.8 percentage points to 98.3%, compared with 91.5% of the original YOLOv5n model. A detection frame rate of 89 frame/s is achieved on the RTX3090 graphics card, which is 1.7 times that of YOLOv5l and 3.7 times that of YOLOX-S, meeting the real-time requirement of TEDS failure detection. The SSD-YOLO algorithm can quickly detect the missing failures of the EMU brake disc bolts, reduces the manual workload of analysts, and provides a reference for future research on condition maintenance of EMU..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0812001 (2025)
Defect Detection of PCB Based on Lightweight ADS-YOLOv8n
Qitao Hu, and Qijie Zou
In view of the issue of balancing detection accuracy with the number of parameters and computational load in printed circuit board (PCB) defect detection, this study proposes a lightweight PCB defect detection algorithm based on ADS-YOLOv8n. Firstly, the ADown downsampling module is introduced to retain more detailed dIn view of the issue of balancing detection accuracy with the number of parameters and computational load in printed circuit board (PCB) defect detection, this study proposes a lightweight PCB defect detection algorithm based on ADS-YOLOv8n. Firstly, the ADown downsampling module is introduced to retain more detailed defect information and enhance the ability to extract detail defects. Secondly, a DTFM module incorporating three layers features is designed to enhance feature extraction and ability to localize defects. Then, a new SCM module is designed to enhance the focus on defect information. Finally, the WIoUv3 bounding box loss function is introduced to enable the model to obtain more accurate regression results. The mean average precision of the improved model reaches 98.43% and the recall rate reaches 96.58%, compared with the benchmark model, the mean average precision is improved by 3.20 percentage points, the recall rate is improved by 5.17 percentage points, and the number of parameters and computation volume are reduced by 5.0×105 and 3.0×108, respectively. The improved model takes into account the lightweight of the model on the basis of improving the detection precision..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0812002 (2025)
Machine Vision
Lidar SLAM Algorithm Based on Point Cloud Geometry and Intensity Information
Long Peng, Weigang Li, Qifeng Wang, and Huan Yi
To address the low accuracy of the laser simultaneous localization and mapping (SLAM) algorithm in large scenes, this paper proposes a lidar SLAM algorithm based on point cloud geometry and intensity information. First, point cloud geometry and intensity information are combined for feature extraction, where weighted iTo address the low accuracy of the laser simultaneous localization and mapping (SLAM) algorithm in large scenes, this paper proposes a lidar SLAM algorithm based on point cloud geometry and intensity information. First, point cloud geometry and intensity information are combined for feature extraction, where weighted intensity is introduced in the calculation of local smoothness to enhance the robustness of feature extraction. Subsequently, a rectangular intensity map is introduced to construct intensity residuals, whereas geometric and strength residuals are integrated to optimize pose estimation, thereby improving mapping accuracy. Finally, a feature descriptor based on point cloud geometry and intensity information is introduced in the loop detection process, thus effectively enhancing the loop recognition accuracy. Experimental results on public datasets show that, compared with the current mainstream LeGO-LOAM algorithm, the proposed algorithm offers greater position accuracy..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0815001 (2025)
Lightweight Small Object Detection Algorithm Based on STD-DETR
Zeyu Yin, Bo Yang, Jinling Chen, Chuangchuang Zhu, Hongli Chen, and Jin Tao
To address the challenges of small target detection in aerial photography images by unmanned aerial vehicle, including complex background, tiny and dense targets, and difficulties in deploying models on mobile devices, this paper proposes an improved lightweight small target detection algorithm based on real-time DEtecTo address the challenges of small target detection in aerial photography images by unmanned aerial vehicle, including complex background, tiny and dense targets, and difficulties in deploying models on mobile devices, this paper proposes an improved lightweight small target detection algorithm based on real-time DEtection TRansformer (RT-DETR) model, named STD-DETR. First, RepConv is introduced to improve the lightweight Starnet network, replacing the original backbone network, thereby achieving lightweight. A novel feature pyramid is then designed, incorporating a 160 pixel × 160 pixel feature map output at the P2 layer to enrich small target information. This approach replaces the traditional method of adding a P2 small target detection head, and introduces the CSP-ommiKernel-squeeze-excitation (COSE) module and space-to-depth (SPD) convolution to enhance the extraction of global features and the fusion of multi-scale features. Finally, pixel intersection over union (PIoU) is used to replace the original model's loss function, calculating IoU at the pixel level to more precisely capture small overlapping regions, reducing the miss rate and improving detection accuracy. Experimental results demonstrate that, compared with baseline model, the STD-DETR model achieves improvements of 1.3 percentage points, 2.2 percentage points, and 2.3 percentage points in accuracy, recall, and mAP50 on the VisDrone2019 dataset, while reducing computational cost and parameters by ~34.0% and ~37.9%, respectively. Generalization tests on the Tinyperson dataset show increases of 3.7 percentage points in accuracy and 3.1 percentage points in mAP50, confirming the model's effectiveness and generalization capability..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0815002 (2025)
DiffSegNet: 3D Point Cloud Instance Segmentation Model Based on Differential 3D U-Net
Gang Yang, Na Zhang, Fan Hu, and Hua Yu
This study addresses the challenge of segmenting instances in substations, where complex shapes and occlusions of various devices hinder edge feature extraction and subsequently reduce segmentation accuracy is low. We propose a 3D instance segmentation model based on differential 3D U-Net. First, we introduce a voxel vThis study addresses the challenge of segmenting instances in substations, where complex shapes and occlusions of various devices hinder edge feature extraction and subsequently reduce segmentation accuracy is low. We propose a 3D instance segmentation model based on differential 3D U-Net. First, we introduce a voxel view enhancement module that enhances voxel features and improves the model's spatial perception by integrating a cross-attention mechanism with double view projection. Second, we construct a high-frequency sensitive differential feature fusion module to enhance the model's ability to learn edge features of instances. Finally, we design a deep fusion module to combine deep semantic features with shallow structural features, thereby enhancing the model's semantic discrimination capabilities. Experimental results demonstrate that the proposed model achieves a mean average precision score of 66.58% under a custom power scenario and 70.29% on the ScanNet dataset, significantly outperforming the mainstream models and showcasing its strong engineering application potential..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0815003 (2025)
3D Reconstruction of Irregular Surface Point Cloud Data for Rail Gates Based on Line Laser
Zelin Liu, Jiaxing Tang, Shengyi Chen, Yexuan Chen, and Tiezheng Guo
A 3D reconstruction method leveraging line laser point cloud processing for complex surfaces is proposed to address the challenges associated with the Fuxing track door, including its irregular surface shape, uneven substrate, and inefficient manual inspection. First, an enhanced image subtraction algorithm utilizing sA 3D reconstruction method leveraging line laser point cloud processing for complex surfaces is proposed to address the challenges associated with the Fuxing track door, including its irregular surface shape, uneven substrate, and inefficient manual inspection. First, an enhanced image subtraction algorithm utilizing sub-pixel extension is designed to increases image resolution 16-fold during line structured light extraction, thereby mitigate detail loss. Second, a point cloud data preprocessing technique that employs combined filtering is introduced to enhance processing efficiency. Finally, a dimensionality reduction and geo-classification simplification algorithm for 3D data is developed to enhance program efficiency while simultaneously extracting valuable edge information. Experimental results demonstrate that the 3D model achieves an average curvature fitting rate of 93%, indicating high precision..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0815004 (2025)
Lossless Point Cloud Encoding Based on Invertible Neural Network
Kaiyuan Wang, Zhijun Fang, and Junxin Lu
Traditional lossless point cloud encoding algorithms suffer from low encoding efficiency, whereas those based on convolutional and autoencoder networks exhibit a certain degree of feature information loss. To address these issues, the study proposes a lossless point cloud encoding algorithm based on invertible neural nTraditional lossless point cloud encoding algorithms suffer from low encoding efficiency, whereas those based on convolutional and autoencoder networks exhibit a certain degree of feature information loss. To address these issues, the study proposes a lossless point cloud encoding algorithm based on invertible neural networks. First, the proposed algorithm leverages mathematically rigorously deducible invertible neural networks to mitigate the loss of feature information during point cloud encoding. Furthermore, a 3D-invertible-block module is introduced to extract global information from the original point cloud, enhancing the accuracy of encoding. Finally, a context occupancy prediction module is deployed to constrain contextual information from the point cloud, enhancing the network's nonlinear expression capability and thereby preserving the complete original point cloud information. Experimental results demonstrate that, compared to the benchmark Geometry-based Point Cloud Compression (G-PCC) method provided by the Moving Picture Experts Group (MPEG), the proposed algorithm exhibits superior encoding performance on the MVUB and MPEG 8i datasets, achieving a 37.25% improvement in terms of the encoding rate..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0815005 (2025)
Improved RRU-Net for Image Splicing Forgery Detection
Ying Ma, Yilihamu Yaermaimaiti, Shuoqi Cheng, and Yazhou Su
To address the problem that feature extraction by increasing the depth in image splicing forgery detection algorithm based on convolutional neural network (CNN) can easily lead to loss of shallow forgery trace features, which causes a decrease in image resolution, this paper proposes an improved ringed residual U-net (To address the problem that feature extraction by increasing the depth in image splicing forgery detection algorithm based on convolutional neural network (CNN) can easily lead to loss of shallow forgery trace features, which causes a decrease in image resolution, this paper proposes an improved ringed residual U-net (RRU-Net) dual-view multiscale image splicing forgery detection algorithm. First, the noise image is generated by multifield fusion, and the noise perspective is generated through the high-pass filter of the spatial rich model (SRM), to enhance edge information learning. Second, the multiscale feature extraction module is designed by combining the original view with continuous downsampling of the noisy view to obtain the multiscale semantic information of the image. Finally, the A2-Nets dual-attention network is introduced to effectively capture the global information and accurately locate the tampered area of ??the image. Compared with the original RRU-Net, the algorithm in this study shows a significant detection effect and robustness improvement on multiple data sets, demonstrating significant progress in the field of image forgery detection. These results show that the proposed method has higher accuracy and reliability when dealing with complex scenes and diversified data, providing important technical support for research and application in the field of image security and information protection..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0815006 (2025)
Voxel RCNN-SLCK 3D Object Detection Algorithm Based on Split Large Convolution Kernels
Guanwu Zhao, and Guansheng Xing
A Voxel RCNN-SLCK 3D object detection model based on split large convolution kernels is proposed to address the issues in current 3D point cloud object detection algorithms, including low detection accuracy for small targets such as pedestrians and occluded targets as well as the tendency to miss or incorrectly detect A Voxel RCNN-SLCK 3D object detection model based on split large convolution kernels is proposed to address the issues in current 3D point cloud object detection algorithms, including low detection accuracy for small targets such as pedestrians and occluded targets as well as the tendency to miss or incorrectly detect these targets. First, an SLCK3D feature extraction module based on split large convolution kernels is designed and embedded into a restructured 3D backbone network to achieve feature extraction capabilities comparable to those realized by using large convolution kernels directly, while reducing the amount of training required. Second, a multi-head self-attention module is added to the 3D backbone network to enhance contextual relationships among features and improve the detection performance for small and occluded targets. Finally, a deformable self-attention (DSA) module is introduced into the 2D backbone network to further enhance adaptability to small and occluded targets, thus enhancing the feature extraction capabilities of the network. Using the KITTI dataset, the proposed algorithm improves the detection accuracy for occluded targets in the categories of cars, pedestrians, and cyclists by 0.82, 1.27, and 0.62 percentage points, respectively, compared with the original algorithm. Experimental results show that the detection performance of the proposed model is significantly improved..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0815007 (2025)
Dynamic SLAM Algorithm Based on Instance Segmentation and Optical Flow Feature Clustering
Heng Zhang, Xiaoqiang Zhang, Guanwu Jiang, Zhixin Zhang, Yang He, and Xuliang Wang
Simultaneous localization and mapping (SLAM) is a technology widely employed in fields such as autonomous driving and augmented reality. Traditional visual SLAM methods assume a static environment, which compromises positioning accuracy and robustness in dynamic settings. To address this limitation, a dynamic SLAM algoSimultaneous localization and mapping (SLAM) is a technology widely employed in fields such as autonomous driving and augmented reality. Traditional visual SLAM methods assume a static environment, which compromises positioning accuracy and robustness in dynamic settings. To address this limitation, a dynamic SLAM algorithm based on instance segmentation and optical flow feature clustering is proposed. A thread for instance segmentation is introduced, utilizing YOLACT real-time instance segmentation to detect potential dynamic targets. Since directly removing the feature points of dynamic targets may result in an insufficient number of static feature points, which adversely affects pose estimation, an optical flow feature clustering algorithm is proposed to further refine the selection of static feature points. In addition, the enhanced and efficient local descriptor BEBLID is employed as a replacement for the original BRIEF feature descriptor to improve matching accuracy. The offline-trained BEBLID bag-of-words model is subsequently utilized to enable relocation and loop detection. Experiments conducted on the TUM dynamic scene dataset demonstrate the effectiveness of the proposed algorithm. When compared with ORB-SLAM2, the algorithm achieves an average reduction of 97.3% in the root mean square error of absolute trajectory error in dynamic environments, with a maximum reduction of 98.1%. Compared with other dynamic SLAM approaches, the proposed method also exhibits improved accuracy to a significant extent..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0815008 (2025)
Three-Dimensional Object-Tracking Method Based on Joint of Multidimensional Features of Point Clouds
Feng Tian, Sirui Zhang, Fang Liu, Zhuohan Han, Mengyang Zhang, Yizhou Lu, Guibao Ma, Huan Chang, Ling Zhao, and Yuxiang Han
Three-dimensional (3D) target tracking is a critical research area in autonomous driving. Most existing methods rely primarily on intersection over union or motion features for data association, whereas the 3D morphological and positional characteristics of the target are typically disregarded, thus resulting in signifThree-dimensional (3D) target tracking is a critical research area in autonomous driving. Most existing methods rely primarily on intersection over union or motion features for data association, whereas the 3D morphological and positional characteristics of the target are typically disregarded, thus resulting in significant matching errors. Hence, we propose a novel 3D target tracking approach that leverages the joint multidimensional features of point clouds. First, an object-embedding matching module developed based on a 3D detection backbone was introduced to extract more discriminative 3D features of the target. Second, a motion prediction module was incorporated, thus enhancing the consistency across frames by utilizing historical trajectory data to forecast the target's future position. Finally, a joint multidimensional affinity matrix was constructed by combining 3D shape features, motion characteristics, and center-point positional information, thereby improving the robustness of trajectory and detection associations. Validation on the public nuScenes dataset demonstrates the superior tracking performance of the proposed method, with a 4.4 percentage points increase in the average multi-object tracking accuracy and a reduction of 322 identity switches, thus significantly mitigating identity switching errors. These results prove the method's enhanced efficacy..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0815009 (2025)
Self-Supervised Monocular Depth Estimation Model Based on Global Information Correlation Under Influence of Local Attention
Lei Xiao, Peng Hu, and Junjie Ma
Current methods for estimating monocular depth based on global attention mechanisms excel in capturing long-range dependencies, however, they often have drawbacks of high computational complexity and numerous parameters. Additionally, these methods can be susceptible to interference from irrelevant regions, which reducCurrent methods for estimating monocular depth based on global attention mechanisms excel in capturing long-range dependencies, however, they often have drawbacks of high computational complexity and numerous parameters. Additionally, these methods can be susceptible to interference from irrelevant regions, which reduces their ability to accurately estimate fine details. This study proposes a self-supervised monocular depth estimation model based on a local attention mechanism, which further leverages convolution and Shuffle operations for global information interaction. The proposed method first calculates attention within divided local windows and then effectively integrates global information by combining depthwise separable convolutions and Shuffle operations across spatial and channel dimensions. Experimental results on the public KITTI dataset demonstrate that the proposed method significantly reduces computational complexity and parameter count and improves the ability to handle depth details, outperforming mainstream methods based on global attention mechanisms..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0815010 (2025)
Small Target Objects and 3D Scene Reconstruction Based on Center Alignment and Improved FPFH
Yige Zhao, Kai Wang, Xiaoke Zhang, Chen Yang, and Hui Chen
Completely reconstructed three-dimensional (3D) scenes are widely used in urban planning, robot mapping, autonomous driving, and augmented reality applications. However, the occlusion of small targets in the scene or issues pertaining to the shooting equipment at a specific angle can easily result in the loss of point-Completely reconstructed three-dimensional (3D) scenes are widely used in urban planning, robot mapping, autonomous driving, and augmented reality applications. However, the occlusion of small targets in the scene or issues pertaining to the shooting equipment at a specific angle can easily result in the loss of point-cloud data. Hence, this paper proposes a method for registering and reconstructing small target objects and 3D scenes based on centroid alignment and improved fast point feature histograms (FPFHs). First, to solve the problem of different data scales for different sensors, a centroid-based transformation method was proposed to scale-align the point clouds of small target objects and the scene. Subsequently, to solve the difficulty in matching feature points between small target objects and the scene, the intrinsic shape signatures (ISS) algorithm was used to extract key points, and the proposed improved FPFH algorithm was used to complete the coarse registration of the point clouds. Finally, the bi-directional iterative closest point (ICP) algorithm was used to complete the fine registration of the point clouds and reconstruct the complete 3D scene. Experimental results show that the proposed method can solve the registration problem between the point clouds of small target objects and the scene point clouds in scenes 1?6 of the self-constructed scene dataset, thus improving the accuracy and completeness of the scene-reconstruction results. Compared with the fast global registration (FGR)+ICP and FPFH+ICP methods, the proposed method offers a higher accuracy by 77.5% on average..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0815011 (2025)
Stereo Matching Algorithm Based on Adaptive Spatial Convolution
Fanna Meng, ZouYongjia, Yang Cao, Jin Lü, and Hongfei Yu
Stereo matching, a significant research focus in computer vision, has wide-ranging applications in fields such as autonomous driving, medical imaging, and robotic navigation. To address the issue of poor matching performance in ill-posed regions within image sequences, this paper presents a stereo matching algorithm baStereo matching, a significant research focus in computer vision, has wide-ranging applications in fields such as autonomous driving, medical imaging, and robotic navigation. To address the issue of poor matching performance in ill-posed regions within image sequences, this paper presents a stereo matching algorithm based on adaptive spatial convolution. Initially, an adaptive spatial convolution block is incorporated into the context network. Through weighted aggregation of multiple convolution kernel responses, the context network's capability to capture pathological regions in complex scenes is enhanced, achieving accurate feature representation. Attention maps of the input features are then obtained along the channel dimension. Subsequently, the algorithm uses a multiscale gated recurrent unit (GRU) network structure to optimize the initial disparity results, and the attention maps generated by adaptive spatial convolution are employed to weight the GRU's output, effectively suppressing noise and further improving the accuracy of disparity estimation. Experimental results show that the proposed algorithm achieves an average endpoint error of 0.46 pixel on the Scene Flow dataset, reducing the error by 14.81% compared to benchmark methods. On the KITTI dataset, it achieves a 3-pixel error rate of 1.40% across all regions, a 15.66% reduction compared to benchmark methods, and delivers superior disparity estimation performance. Notably, in ill-posed scenarios such as occluded or reflective regions, the algorithm effectively retains detailed image features..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0815012 (2025)
Point Cloud Registration and Modeling Method for Gear Surfaces Based on Laser and Vision Fusion
Xingbang Zhao, Zhengminqing Li, Xiaofeng Yu, Yong Liu, and Letian Li
Aiming to address the problems of poor accuracy and low efficiency in modeling gear surfaces, as well as the lack of surface detail information in three-dimensional (3D) reconstruction models of laser point clouds, a point cloud registration and modeling method for gear surfaces based on laser and vision fusion is propAiming to address the problems of poor accuracy and low efficiency in modeling gear surfaces, as well as the lack of surface detail information in three-dimensional (3D) reconstruction models of laser point clouds, a point cloud registration and modeling method for gear surfaces based on laser and vision fusion is proposed. First, considering the high similarity between 3D features of the complete point cloud of the gear surface and the point cloud of the registration overlap area, a coarse registration method suitable for point clouds of gear surfaces is proposed. Second, aiming to address the issues of low convergence speed and poor accuracy of the iterative closest point (ICP) algorithm, improvements are made to the ICP algorithm by introducing curvature and voxel grid filtering to realize point cloud fine registration. Finally, to address the lack of surface detail information, such as color and texture, in 3D reconstruction models of laser point clouds, the checkerboard calibration method and the direct linear transform (DLT) method are combined to solve for the camera pose parameters . The laser point cloud is colored by the point cloud library (PCL) to realize the realistic modeling of gear surfaces. The experimental results indicate that compared with the traditional registration method, the proposed method reduces the root mean square error by 61% and the registration time by 53%. Unlike laser point cloud modeling, the proposed 3D reconstruction model incorporates surface texture information..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0815013 (2025)
Medical Optics and Biotechnology
Low-Dose CT Image Denoising Based on Error Modulation Module
Xiaohe Zhao, Ping Chen, and Jinxiao Pan
To address the issue of degraded image quality in low-dose X-ray computed tomography (LDCT) caused by significantly reduced ionizing radiation doses in current denoising methods, this paper proposes an enhanced diffusion model based on the U-Net network. The model introduces an error modulation module to resolve the prTo address the issue of degraded image quality in low-dose X-ray computed tomography (LDCT) caused by significantly reduced ionizing radiation doses in current denoising methods, this paper proposes an enhanced diffusion model based on the U-Net network. The model introduces an error modulation module to resolve the problem of error accumulation during sampling. Additionally, a composite loss function that combines
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0817001 (2025)
Simulation Design and Optical Signal Optimization of Blood Oxygen Sensor Based on TracePro
Sizhe Ye, Hongyi Yang, En Ma, and Fulin Lin
As an important component of wearable devices, the key parameters of the blood oxygen sensor (detection efficiency, uniformity, etc.) directly affect the performance indicators of the device. Based on the TracePro simulation platform, the simulation design of blood oxygen sensor is carried out to optimize the light proAs an important component of wearable devices, the key parameters of the blood oxygen sensor (detection efficiency, uniformity, etc.) directly affect the performance indicators of the device. Based on the TracePro simulation platform, the simulation design of blood oxygen sensor is carried out to optimize the light propagation signal. A multilayer skin tissue model conforming to Beer-Lambert's law and Henyey-Greenstein scattering function is constructed to explore the propagation law and difference of light in various skin tissues under the influence of absorption and scattering. For the blood oxygen sensor module, the spectrum and spatial light distribution of the light source and the spectral-angular response characteristics of the detector are simulated. Multi-channel module design is carried out, and the design of detector space layout, detection distance, center distance, light window, and anti-spurious light structure are optimized with luminous flux and illumination uniformity as evaluation indicators. The surfaces of constant-angle and variable-angle Fresnel lenses are constructed, the influences of surface parameters for Fresnel lenses on the spatial light distribution are compared, and the optical signal transmission before and after optimization is compared by the variable-angle Fresnel lens optimization module. Results show that the detection efficiency is increased by up to ~27% and the illumination uniformity is increased to 94.45% after optimization..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0817002 (2025)
Application of Multi-Level Feature Fusion Method Combined with Transformer in Brain Tumor Diagnosis
Yang Bai, Dejian Wei, Boru Fang, Liang Jiang, and Hui Cao
As a significant brain disorder, brain tumors constantly threaten human health, making early diagnosis crucial. In the analysis of existing methods, deep learning, with its ability to automatically extract features through multi-level nonlinear transformations, demonstrates superior performance in diagnosing brain tumoAs a significant brain disorder, brain tumors constantly threaten human health, making early diagnosis crucial. In the analysis of existing methods, deep learning, with its ability to automatically extract features through multi-level nonlinear transformations, demonstrates superior performance in diagnosing brain tumors. In this study, a deep-learning-based brain tumor diagnosis model that focuses on multi-level feature analysis is proposed. ResNext is employed to extract multi-level features. A spatial attention mechanism combining linear layers and large-kernel convolutions is designed to analyze multi-level contextual information. Moreover, the Transformer structure is integrated to dynamically fuse multi-level features, generating feature maps with high expression power for the final diagnosis. The model is trained and evaluated on the Kaggle dataset for two-class and four-class brain tumor classification tasks. Experimental results show that the model achieves an accuracy of 99.47% in distinguishing between no tumor and tumor, and an accuracy of 99.75% in distinguishing between no tumor and three types of tumors. Compared with other deep-learning models, the proposed method demonstrates superior diagnostic capabilities, enabling early diagnosis with high accuracy..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0817003 (2025)
Optics in Computing
Image Detection Algorithm for Optical Data Storage Based on Sobel Filtering
Ziquan Wang, Bo Zhang, Zhuo Wang, and Jianrong Qiu
With the vigorous development of new-generation information technologies, such as artificial intelligence and 5G communication, the efficient and low-energy storage of large-scale data have become a research hotspot. Optical storage has been extensively studied due to its characteristics such as large storage capacity With the vigorous development of new-generation information technologies, such as artificial intelligence and 5G communication, the efficient and low-energy storage of large-scale data have become a research hotspot. Optical storage has been extensively studied due to its characteristics such as large storage capacity and long lifespan. The algorithm used for data extraction is vital for data storage because it directly affects the bit error rate and recognition speed in data storage. To further improve the accuracy and speed of optical storage data reading, a multidimensional optical storage data reading image detection algorithm based on Sobel filtering is proposed in this study. Photos are preprocessed using Sobel and 2D filterings to suppress interlayer crosstalk between adjacent data layers in multidimensional optical storage. A lattice grid is drawn using connected regions and classification algorithms, and data are identified by analyzing the complex image features of a single grid. Experimental results show that this algorithm reduces the error rate of multidimensional optical storage to 0.3% and improves the data reading accuracy and recognition speed. Furthermore, it possesses important application value in the fields of optical storage and image recognition..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0820001 (2025)
Remote Sensing and Sensors
High-Resolution Remote-Sensing Change Detection Algorithm Based on Contextual Siam-UNet++
Jian Zheng, and Zihang Xu
To address common challenges in change detection tasks, such as the loss of contextual feature information and insufficient utilization of data features that exploit only single-channel or spatial information, a high-resolution remote-sensing change detection algorithm based on context-aware Siam-UNet++ is proposed. ThTo address common challenges in change detection tasks, such as the loss of contextual feature information and insufficient utilization of data features that exploit only single-channel or spatial information, a high-resolution remote-sensing change detection algorithm based on context-aware Siam-UNet++ is proposed. This algorithm uses UNet++ as its backbone network and incorporates a transformer-style contextual transformer module to obtain contextual information from bitemporal images. Hence, more precise image change features are obtained. In addition, an ensemble attention module is employed to comprehensively utilize both channel and spatial information, which leads to higher precision and accuracy in change detection tasks. Experimental validation on the LEVIR-CD and WHU-CD datasets yields F1 scores of 90.07% and 92.19% as well as intersection over union scores of 81.93% and 85.51%, respectively. Experimental results confirm the effective enhancement of the detection performance and accuracy of proposed algorithm..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0828001 (2025)
Multi-View Three-Dimensional Reconstruction of Weak Texture Regions Based on Simple Lidar
Tingya Liang, Xie Han, and Haoyang Li
To address the issues of holes and regional deficiencies in weakly textured areas during multi-view stereo (MVS) three-dimensional (3D) reconstruction, this paper proposes a complementary 3D reconstruction method that fuses data from MVS and lidar. First, the GeoMVSNet deep learning network is used to process multi-vieTo address the issues of holes and regional deficiencies in weakly textured areas during multi-view stereo (MVS) three-dimensional (3D) reconstruction, this paper proposes a complementary 3D reconstruction method that fuses data from MVS and lidar. First, the GeoMVSNet deep learning network is used to process multi-view images captured by a smartphone camera, resulting in MVS depth maps. Next, the smartphone camera and lidar are calibrated to compute the internal and external parameters and the coordinate system transformation matrix. Through transformations of temporal and spatial consistency, viewing angles, and scale, the sparse point cloud collected by lidar is converted to the image perspective. In addition, a depth map enhancement algorithm is proposed and applied to generate dense point cloud data from the lidar. Finally, the dense depth maps generated by lidar and the MVS depth maps are fused. Experimental results show that the reconstruction quality is significantly improved in weakly textured areas on a self-built dataset when the proposed method is used, enhancing the accuracy and completeness of 3D reconstruction. Thus, this study provides an effective solution for 3D reconstruction in weakly textured regions..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0828002 (2025)
Improved ResUNet Method for Extracting Buildings from Remote Sensing Images
Qianrong Sun, and Xiaopeng Wang
To address the limitations in existing semantic segmentation networks for remote sensing images, specifically the loss of fine details, insufficient focus on local information, and restricted ability to capture multi-scale contextual information, SGMFResUNet, a deep residual network enhanced by a spatial information-enTo address the limitations in existing semantic segmentation networks for remote sensing images, specifically the loss of fine details, insufficient focus on local information, and restricted ability to capture multi-scale contextual information, SGMFResUNet, a deep residual network enhanced by a spatial information-enhanced global attention mechanism and multi-module scale fusion for automatic building features extraction from remote sensing images, is proposed in this study. Built upon the ResUNet architecture, the proposed model increases network depth to capture richer multi-level features and employs asymmetric convolutional blocks to improve feature representation and extraction capabilities at various levels. A dual pooling dense pyramid module is designed to enable dense capture of multi-scale contextual information, supplemented by global features. In addition, a hierarchical detail enhancement module is designed to progressively integrate shallow-level features, reducing detail loss. The improved global attention mechanism further adapts feature adjustment, enhancing cross-dimensional interactions to provide more robust feature representation. Experiments on the WHU and Massachusetts building datasets demonstrate that SGMFResUNet achieves intersection over union scores of 90.74% and 77.00%, representing improvements of 1.93 percentage points and 3.29 percentage points, respectively, with respect to ResUNet. Furthermore, compared to ResUNet, HRNetV2, MSFCN, BuildFormer, DC-Swin, and SDSC-UNet, SGMFResUNet consistently demonstrates superior accuracy in building extraction..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0828003 (2025)
Research on Single-Photon Sparse Point Cloud Spatio-Temporal Correlation Filtering Algorithm
Zeyu Guo, Zhen Chen, Bo Liu, Enhai Liu, and Huachuang Wang
Long-distance noncooperative target ranging echo point cloud acquired by single-photon LiDAR is sparse and contains a significant amount of noise, which renders it difficult to accurately extract effective echo and distance trajectory in real time. Therefore, a real-time extraction algorithm for single-photon LiDAR ranLong-distance noncooperative target ranging echo point cloud acquired by single-photon LiDAR is sparse and contains a significant amount of noise, which renders it difficult to accurately extract effective echo and distance trajectory in real time. Therefore, a real-time extraction algorithm for single-photon LiDAR ranging trajectory is proposed based on its temporal and spatial correlation. First, the distance-trajectory-extraction problem is converted into a curve-recognition problem based on the Hough transform, and sparse echo point cloud data is extracted from point-cloud data containing a significant amount of noise by adaptively optimizing the Hough transform. Subsequently, the extracted echo point cloud is used as the observation value of Kalman filtering to accurately estimate the target distance trajectory. In frame segments in which the target is occluded or noise exerts a significant effect, thus resulting in frame segments with missing targets owing to occlusion or severe noise, the distance and velocity information of the target in the previous frame is utilized to predict the state of the target in the current frame. The proposed algorithm processes the simulated data with a minimum signal-to-noise ratio of -9.42 dB under extreme motion conditions. Additionally, the smallest root-mean-square error of the target's distance is 0.833 m, and the lowest leakage-detection rate is 0.05. Moreover, the algorithm offers better real-time performance and can output the target distance continuously and accurately at a frame rate of 100 Hz. During the processing of the measured data, when the echo missed 885 frames continuously, the distance trajectory predicted by the algorithm deviates from the actual value by only 0.2 m. Thus, the proposed algorithm provides an effective method for the real-time accurate extraction of single-photon LiDAR ranging trajectory..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0828004 (2025)
Reviews
Application Prospects of Iris Recognition Technology in Identity Authentication for Standalone Virtual Reality Devices
Zhenyu Ma, Nini Wang, Guolei Wu, Chenlong Zhu, Mingqi Zhou, Yongjun Liu, and Yunxiang Yan
With the iteration and advancement of underlying technologies such as 5G, artificial intelligence, and blockchain, virtual reality (VR) technology is steadily maturing and integrating into a wide range of industries, including party-building training, landscaping, social gaming, and more. In the future, VR devices may With the iteration and advancement of underlying technologies such as 5G, artificial intelligence, and blockchain, virtual reality (VR) technology is steadily maturing and integrating into a wide range of industries, including party-building training, landscaping, social gaming, and more. In the future, VR devices may become essential tools in people's daily lives, work environments, and educational contexts. Among these consumer electronic products, identity authentication plays a pivotal role. Secure and efficient identity authentication mechanisms can safeguard consumers' personal information and digital assets while enhancing the overall human-machine interaction experience. However, current user authentication methods for VR products are often cumbersome, posing significant usability challenges.This paper introduces, the application of iris recognition technology to standalone VR devices, aiming to achieve secure and convenient identity authentication. The discussion begins by evaluating the advantages and limitations of existing authentication methods, such as fingerprint recognition, facial recognition, voice recognition, and iris recognition. It further analyzes the feasibility of incorporating iris recognition technology into standalone VR devices for identity authentication. Subsequently, the paper elaborates on the hardware requirements and implementation strategies for integrating iris recognition functionality into standalone VR devices. Lastly, it addresses the technical challenges associated with this implementation and outlines the future developmental directions for applying iris recognition technology in identity authentication for standalone VR devices..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0800001 (2025)
Application of Liquid Lens in Microscopic Imaging System
Haoran Zhang, Zhengchao Wang, Yi Zheng, Youran Zhao, Chao Liu, and Qionghua Wang
Liquid lens is an optical device with lens function formed by liquid filling, which changes the curvature of the liquid surface through electronic control drive, pressure drive and other methods, thereby changing its focal length. As an emerging photonic device, liquid lenses have unique advantages such as strong tunabLiquid lens is an optical device with lens function formed by liquid filling, which changes the curvature of the liquid surface through electronic control drive, pressure drive and other methods, thereby changing its focal length. As an emerging photonic device, liquid lenses have unique advantages such as strong tunability, fast response speed, miniaturization, and low power consumption, providing a solution for designing high-performance microscopy imaging systems. This article provides an overview of the typical technologies and application progress of liquid lenses in microscopy imaging systems. Starting from three aspects—continuous zoom, adjustable depth of field, and miniaturization, it introduces the characteristics and advantages of liquid-lens-based microscopy imaging systems, analyzes the limitations and current solutions of liquid lens applications in microscopy imaging systems, and finally summarizes the application of liquid lenses in microscopy imaging systems..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0800002 (2025)
Dark Field Light Scattering Imaging Coupled with Surface Enhanced Raman Scattering for Analysis and Detection Application
Xinyu Lan, Guojun Weng, Xin Li, Jianjun Li, Jian Zhu, and Junwu Zhao
Dark field light scattering imaging is a kind of scattered light microscope imaging technique that produces a dark background by obliquely illuminating the sample with incident light. It has the characteristics of low background signal and high signal-to-noise ratio, but it is difficult to identify material compositionDark field light scattering imaging is a kind of scattered light microscope imaging technique that produces a dark background by obliquely illuminating the sample with incident light. It has the characteristics of low background signal and high signal-to-noise ratio, but it is difficult to identify material composition and internal component. Surface enhanced Raman scattering spectrum is a fingerprint spectrum of molecular vibrations with the characteristics of high sensitivity and non-destructive testing. Noble metal nanoparticles are usually used as the substrates for signal enhancement, but there is no clear standard for the selection of Raman "hot spot" area of micro-measurement, resulting in large fluctuations in Raman signals. The combination of dark field light scattering imaging and surface enhanced Raman scattering allows not only high-resolution microscopic localization of Raman "hot spot" region, but also highly sensitive detection of components in dark field imaging, and high-resolution detection of sample morphology and components can be realized simultaneously in time and space. In this review, the advantages, principles, and applications of dark field light scattering imaging coupled with surface enhanced Raman scattering are summarized, and the recent applications in biochemical molecular detection, cell detection, chemical reaction dynamic monitoring, and other aspects are introduced, which provide new ways for biomedical detections..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0800003 (2025)
Advances in Linear-Wavenumber Spectral Domain Optical Coherence Tomography
Yuxuan Li, Yanping Huang, Jingjiang Xu, Jia Qin, Lin An, and Gongpu Lan
Spectral domain optical coherence tomography (SD-OCT) has achieved significant progress in biomedical imaging, particularly in ophthalmology. However, SD-OCT exhibits reduced signal sensitivity in deeper imaging regions compared to swept-source OCT due to the phenomenon of sensitivity roll-off. Linear-wavenumber spectrSpectral domain optical coherence tomography (SD-OCT) has achieved significant progress in biomedical imaging, particularly in ophthalmology. However, SD-OCT exhibits reduced signal sensitivity in deeper imaging regions compared to swept-source OCT due to the phenomenon of sensitivity roll-off. Linear-wavenumber spectroscopy employs an optical solution to achieve linear spectral splitting in the wavenumber domain, thereby enhancing the system's signal-to-noise ratio in the depth direction. This paper presents a comprehensive overview of the fundamental principles of the SD-OCT system, the spectroscopic theory of linear-wavenumber, the optical model of linear-wavenumber dispersion, and the application of linear-wavenumber SD-OCT in the imaging of soft tissue structures, angiography, and elastography. Although the application market for linear-wavenumber OCT remains relatively small, its ability to improve the signal-to-noise ratio and imaging speed is noteworthy. As the technology continues to advance, linear-wavenumber SD-OCT is expected to play an increasingly significant role in clinical diagnosis and biomedical research..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0800004 (2025)
Spectroscopy
Multispectral Data Correction Method for Painted Cultural Relics Under Non-Uniform Illumination Based on Prior Feature Constraints
Jiachen Li, Ke Wang, Huiqin Wang, Zhan Wang, and Peize Han
To address the problem of spectral information shift caused by uneven illumination during multispectral imaging of painted cultural relics in complex environments, a region-adaptive multispectral image correction model constrained by a priori spectral features is proposed. First, multiscale guided filtering is used to To address the problem of spectral information shift caused by uneven illumination during multispectral imaging of painted cultural relics in complex environments, a region-adaptive multispectral image correction model constrained by a priori spectral features is proposed. First, multiscale guided filtering is used to estimate the illumination component, effectively addressing issues of edge blurring and halo artifacts. The image is then divided into regions, and an adaptive gamma correction function is developed based on the illumination characteristics of these regions to process the illumination component, and balance bright and dark areas. This process is combined with contrast-limited adaptive histogram equalization to enhance local contrast, and the result is fused with the original illumination component. Finally, spectral correlation and consistency across multiple imaging channels are constrained by introducing spectral structure and gradient loss functions as a priori features, characterizing the spatial structure and trend variations of the multispectral images. Experimental results show that compared with other algorithms, the regional adaptive illumination correction algorithm constrained by a priori features reduces the root-mean-square error (RMSE) by an average of 26.96% and increases the spectral correlation measure (SCM) by 21.63% for simulated murals. For real murals, the RMSE is reduced by 11.12% on average, while the SCM is improved by 12.65%, significantly enhancing pigment classification accuracy..
Laser & Optoelectronics Progress
- Publication Date: Apr. 25, 2025
- Vol. 62, Issue 8, 0830001 (2025)
© Copyright 2018-2021 | Chinese Laser Press. All Rights Reserved 沪ICP备15018463号-20