Research on evaluation method of stereo vision measurement system based on parameter-driven

Abstract Stereo vision measurement system is widely used in aerospace manufacturing, intelligent robot and other fields because of its unique advantages. However, when using the stereo vision measurement system for measurement, there is no guidance for the parameter configuration of the measurement system for the scene and task constraints, resulting in the stereo vision measurement system cannot play the maximum performance. In this paper, from two aspects of local measurement of camera parameters and global measurement of system parameters, the relevant model is established to analyze the measurement performance of a single independent camera and the performance of the whole stereo vision measurement system. Firstly, the relationship between single camera calibration parameters and measurement accuracy is analyzed; secondly, the relationship between stereo measurement system parameters and measurement accuracy is analyzed; finally, the stereo vision measurement error function is established to guide the design of measurement system parameters. According to the actual measurement experiment, the correctness of the analysis and the accuracy of the established function are verified, it is efficiently used for guiding the parameter design and camera selection of the measurement system in different measurement scene and task.