Polarization detection technology plays an important role in fields such as environmental exploration, remote sensing imaging, materials science, biomedical applications, and aerospace. Compared with traditional light intensity and spectral detection technologies, polarization detection can provide information about the target's material, refractive index, surface normal direction, etc., significantly improving the capability of obtaining and analyzing target information. However, traditional polarization detection systems are large in size, slow in detection, and difficult to meet the requirements for miniaturization, light weight, and rapid detection. In recent years, the development of metasurface technology has brought new opportunities for polarization detection. Metasurface devices can flexibly control the phase, amplitude, and polarization of light, providing effective solutions for miniaturized and lightweight polarization detection devices. In recent years, the Metagrating designed by Capasso team which achieves full-Stokes vector polarization detection and imaging by separate or control polarization information in the same region, making the device smaller and easier to integrate. Furthermore, by applying Fourier optical theory, the Metagrating achieves precise optimization of the size of each nanocolumn, avoiding cross-interference and losses caused by light field coupling between multiple regions, thus providing high polarization detection energy efficiency. However, due to factors such as processing precision and system integration errors, there is a deviation between the theoretical and actual instrument matrix value of the Metagrating, leading to the decrease of polarization detection accuracy. Therefore, the instrument matrix must be calibrated before polarization detection. The commonly used calibration method (two-step method) for the instrument matrix contains linearly polarized light and circularly polarized light calibration, respectively. The calibration procedure is cumbersome and time-consuming. Additionally, the calibration accuracy is also affected by the imperfect circularly polarized light in lab condition. Accumulated errors from multiple calibration experiments will further reduce the accuracy of polarization detection. To address the disadvantages of traditional instrument matrix calibration methods, this paper proposes a “one-step” calibration method based on elliptical polarized light, which only requires one experiment and avoids errors caused by imperfect circularly polarized light, effectively simplifying the calibration process and avoiding the accumulation of measurement errors.This paper uses a two-dimensional Metagrating with four independent polarization analysis channels (order (1, 0), order (-1, 0), order (0, 1) and order (0, -1)). When light with any polarization state is incident on the Metagrating, diffraction light with specific polarization is generated in the four polarization analysis channels. This modulation process is represented. During polarization detection, four light intensities from the polarization analysis channels can be simultaneously obtained by the detector. If the instrument matrix A is known, the Stokes vector of the incident light can be determined, achieving polarization detection. This process is expressed, where the instrument matrix A of the Metagrating is a 4×4 matrix. When use two-step method to calibrate linear polarization component, S₄=0, so the fourth column circular polarization component of the instrument matrix cannot be calibrated. However, if all four parameters of the Stokes vector are non-zero and their values change with the incident polarization angle, the four parameters of the instrument matrix can be calibrated simultaneously in a single experiment. Considering the four Stokes parameters of elliptical polarized light are non-zero, we propose a one-step calibration method based on elliptical polarized light to simplify the calibration process.Experiments comparing the effects of the “one-step” method and the “two-step” method on the polarization detection accuracy of 1 550 nm two-dimensional Metagrating are shown in Figures 3 to 5. As shown in Table 1, it can be seen that, although the Root Mean Square Error (RMSE) for Degree of Polarization (DOP) of linearly polarized light using the “one-step” method is slightly higher than that of the “two-step” method, the magnitude is the same. The RMSE for the Angle of Polarization (AOP) increases by less than 1°, with only a 0.45% decrease in accuracy. In elliptical polarized light detection, the RMSE values for AOP, Ellipticity of Polarization (EOP), and DOP using the “one-step” method decrease by more than 2.5°, 2.0°, and 0.016, respectively, improving the detection accuracy by 3.05%, 2.76%, and 1.65%. In partial polarization DOP detection, the RMSE for the “one-step” method is also smaller, with accuracy improving by 0.51%. To validate the robustness of the “one-step” calibration approach across different experimental platforms, we performed analogous experiments using a 500 nm Metagrating. It shows similar trends to the 1 550 nm experimental results. This method significantly improves the detection accuracy of elliptical polarized light and partial polarized light, with the RMSE for AOP, EOP, and DOP decreasing by 2.902°, 1.883°, and 0.020, respectively, and detection accuracy improving by 2.66%, 2.16%, and 2.08%. The above results demonstrate that the “one-step” method effectively simplifies the instrument calibration steps before polarization detection, significantly improving the accuracy of polarization Metagrating detection, and provides a new technical approach for constructing high-precision, fast polarization detection systems.