Design and Implementation of Machine Vision Inspection System for Micro Gear

In order to ensure the accuracy of the calculation and quantification results of tooth length error, a machine vision detection system with perfect functions and strong practicability is designed. Firstly, the wavelet transform method is used to completely remove the image noise, and the Radom transform algorithm is used to complete the accurate and scientific correction of the part image. Then, the cubic curve model is used to complete the comprehensive and complete extraction of sub-pixel edge information, and then; The projection mapping method is used to calculate the boundary position accurately. Finally, the dynamic range of the gear center point is calculated accurately, and according to the final calculation result, whether the tooth length error calculation is reasonable or not is accurately judged. The experimental results show that the precision and accuracy of the machine vision inspection system designed in this paper can reach 2μ m and 99% respectively. In addition, the inspection time of a single frame and a single part is relatively short, which is 18ms and 5s, respectively. In short, the system has the characteristics of high inspection efficiency and accuracy, which fully meets the relevant standards and requirements of industrial inspection.