A Components Measuring System Based on Computer Vision
Minghao Zhan, Xiaonan Hou, Xinran Wang, Chunlei Wang · 2022
Components with high precision fit need to be measured after machining. Generally, Coordinate Measuring machines (CMM) are used for those high precision measurement. However, it’s a huge cost of time to proceed high precision measurement with systems like CMM. And they inevitably require manual operation and false detection might occur due to human factors like fatigue. In this paper a components measuring system based on computer vision is proposed to achieve batch inspection and measurement of industrial components. Image processing (CV) part of this system is the focus of this paper while the mechanical part is ignored. Generally, a series of image processing is executed to get the measuring result step by step. Image processing algorithm flow includes image pre-processing, splicing, feature alignment, edge detection and Hough transformation. The experiment at last shows that the proposed system in this paper has nearly the same accuracy and a higher efficiency compared to a measuring process with traditional high precision system CMM. The average deviation of the proposed system is under 10 microns over the accuracy tolerance of 20 microns, and the measuring time cost reduced to 2.6 seconds per ROI.