Upconversion infrared (IR) detection, a method leveraging nonlinear parametric upconversion, effectively upconverts the signal beam spectrum from the IR to the visible band, thus enabling the indirect detection of targets using silicon-based detectors or cameras. However, like most traditional imaging technologies, in practical applications, the scattering and reflection of light caused by fine particles can lead to issues such as gray images and information loss. This limits the application range of this technology. Polarization imaging detection can extract the polarization of the target based on the polarization difference between the target and background light. This allows for obtaining more information independent of intensity and spectrum, thus avoiding the shortcomings of traditional detection imaging technology. It also helps weaken background noise, improve detection and recognition ability, and has a wide range of applications in target detection classification, three-dimensional target reconstruction, and biomedicine. Therefore, combining upconversion with polarization detection technology is an important technical means to accelerate the development of infrared detection technology. This study proposes and experimentally validates a time-division multiplexed nonlinear Stokes tomography scheme based on an electrically controlled Variable Retarder (VR), with a focus on the structural design of a polarization-resolved upconversion detection system. The proposed system functions as a time-division Stokes upconversion tomography device, operating through a cyclic process involving "VR delay control—signal triggering—data acquisition." The electrically controlled liquid crystal VR, with millisecond-level response time, sequentially transmits the different polarization Stokes components of the signal light field into a bulk quasi-phase matchings crystal for efficient frequency upconversion. Ultimately, a visible array detector performs time-division sampling to reconstruct the polarization image. This scheme does not require a complex opto-mechanical structure, involves no mechanical components, and fully utilizes the spatial resolution of visible detectors. The study offers a practical and innovative technological pathway for research related to infrared polarization-resolved imaging techniques.A time-division multiplexed polarization-resolved upconversion detection system is implemented based on VR, combined with spatial Stokes tomography and upconversion detection technology (Fig.1(a)). The system is capable of collecting each Stokes component of the signal beam in real-time and obtaining the spatial polarization distribution of the signal beam after Gaussian profile correction (Fig.1(b)). Additionally, a set of vector beam polarization detection optical paths is constructed to verify the detection effect of this system (Fig.2).This study first use vectorial Ince-Gaussian (IG) modes at different ellipticities ($ \varepsilon $) as detection targets (Fig.3). In particular, as $ \varepsilon =0 $ and $ \epsilon \longrightarrow {\infty } $, the IG modes converted to Laguerre-Gaussian (LG) and Hermite-Gaussian (HG) modes of the same modal orders, respectively. The experimental results show that the IG vector modes detected using the time-division multiplexed nonlinear Stokes tomography principle exhibit an excellent match with the direct observation results obtained through an InGaAs polarization camera at the 1560 nm wavelength. Based on the Gaussian profile correction Stokes components, the spin-orbit coupling (SOC) states and spatial polarization distributions of the relevant target vector modes can be reconstructed. By comparing the Signal and the Reconstruction in the SOC states on the SOC sphere, both the spatial polarization structure and the position on the SOC spatial mode sphere align with high fidelity, with all fidelity metrics exceeding 99%. Building on this, the performance of the experimental system was further validated using more generalized polarization images carried by non-paraxial beams (Fig.4). The detection results indicate that the spatial polarization distribution of the reconstructed upconverted beam aligns well with that of the preset images.This paper addresses the structural complexity of infrared polarization upconversion detection technology by proposing a novel method based on electrically controlled VR, termed "Time-Division Multiplexed Nonlinear Stokes Tomography." The proposed approach utilizes an electrically controlled VR loop to achieve millisecond-level time-response, enabling time-division upconversion and capturing the polarization components of the target beam to reconstruct the polarization image. Quantitative analysis of the detection results for vector modes demonstrates that the upconverted light field can faithfully inherit the spatial polarization distribution of the signal light field. Furthermore, by targeting generalized polarization images carried by non-paraxial Gaussian beams, the potential of this system for practical applications is explored. The study shows that this method can achieve high-precision tomographic detection of the target polarization state while simultaneously accomplishing spectral migration. The proposed method features a compact structure and strong robustness, offering a new technical route for polarization imaging and detection in the infrared or terahertz bands, with broad application prospects in fields such as astronomical observation, disease diagnosis, and environmental monitoring.