Improved Maize Plant Contour Extraction Method

Jinxin Liang; Le Zhang; Yuyao Meng; Jie Teng; Quanling He; Leiyang Fu; Shaowen Li

doi:10.3788/LOP220561

Journals >Laser & Optoelectronics Progress >Volume 60 >Issue 12 >Page 1210004 > Article

Laser & Optoelectronics Progress
Vol. 60, Issue 12, 1210004 (2023)

Improved Maize Plant Contour Extraction Method

Jinxin Liang^1,2, Le Zhang³, Yuyao Meng^1,2, Jie Teng^1,2..., Quanling He^1,2, Leiyang Fu^1,2 and Shaowen Li^1,2,*|Show fewer author(s)

Author Affiliations

¹School of Information & Computer, Anhui Agricultural University, Hefei 230036, Anhui, China

²Anhui Provincial Key Laboratory of Smart Agricultural Technology and Equipment, Hefei 230036, Anhui, China

³Anhui Electric Power Transmission and Transformation Co. Ltd., Hefei 230036, Anhui, China

show less

DOI: 10.3788/LOP220561 Cite this Article Set citation alerts

Jinxin Liang, Le Zhang, Yuyao Meng, Jie Teng, Quanling He, Leiyang Fu, Shaowen Li. Improved Maize Plant Contour Extraction Method[J]. Laser & Optoelectronics Progress, 2023, 60(12): 1210004 Copy Citation Text

show less

Fig. 1. Flow chart of maize contour extraction

Download full size

Fig. 2. Robot experimental platform

Download full size

Fig. 3. Experimental maize plant images. (a) One bud with three leaves 30°; (b) one bud with three leaves 45°; (c) one bud with three leaves 45°; (d) one bud with four leaves 15°

Download full size

Fig. 4. Comparison of different algorithms

Download full size

Fig. 5. Effect of each contour extraction algorithm. (a)HSV; (b) proposed method; (c) HSV after noise reduction; (d) proposed method after noise reduction

Download full size

Algorithm 1 RANSAC

Inputs：Dataset K

1：Initialize dataset K

2：Initialize expected iterations N

3：for N from 0 to n-1 do

4： Repeat

5： Randomly select m samples from Dataset K for model fitting

6： Selected range ɛ as the fitting interval

7： Count the number $N_{n u m}$ of points in interval ɛ

8： If

9： $N_{n u m}$ greater than an acceptable threshold t，refitting the data sample points

10： Otherwise

11： Set the current model M as the best model

12： End algorithm

13： Until n is not less than the present termination times

14：Set model M in the model set $M_{n}$ ，which with the smallest fitting error in the previous step as the best model

Output：Model M

Table 0. [in Chinese]

Algorithm	Time /s	Keypoints 1	Keypoints 2	$N_{c}$ / $N_{s}$	$R_{c}$ /%
SIFT	1.293	170	116	11/50	22.00
SURF	1.845	660	238	28/50	56.00
ORB	0.638	794	609	32/50	64.00
F-B	0.534	384	282	42/50	84.00

Table 1. Comparison of the time, the number of feature points, the number of matching, and the accuracy of each algorithm

Algorithm	Time /s	Keypoints 1	Keypoints 2	$N_{c}$ / $N_{s}$	$R_{c}$ /%
SIFT	2.228	308	87	23/50	46.00
SURF	3.912	2276	957	39/50	78.00
ORB	1.157	800	800	26/50	52.00
F-B	0.802	818	585	43/50	86.00

Table 2. Comparison of the time, the number of feature points, the number of matching, and the accuracy of each algorithm

Algorithm	Canny	Sobel	Laplacian	Roberts	Prewitt
$R_{p}$	2.01	7.18	2.29	6.49	7.47
$R_{H S V}$	1.33	6.44	2.13	5.55	6.11
$C_{d}$	0.68	0.74	0.16	0.94	1.36

Table 3. Proportion and difference of different methods in each contour extraction algorithm

Algorithm	Canny	Sobel	Laplacian	Roberts	Prewitt
$R_{p}$	1.53	7.95	2.40	7.09	7.89
$R_{H S V}$	1.30	7.71	2.15	6.82	7.40
$C_{d}$	0.23	0.24	0.25	0.27	0.49

Table 4. Proportion and difference of different methods in each contour extraction algorithm 2

Jinxin Liang, Le Zhang, Yuyao Meng, Jie Teng, Quanling He, Leiyang Fu, Shaowen Li. Improved Maize Plant Contour Extraction Method[J]. Laser & Optoelectronics Progress, 2023, 60(12): 1210004

Download Citation

Set citation alerts for the article

Tools

Set citation alerts for the article

Save the article for my favorites

Paper Information

微信扫一扫：分享

微信扫一扫：分享