In order to solve the problem that the size of yoke iron is difficult to measure in the electromagnetic system of the 1 /2 crystal cover relay, a high-precision automatic detection system of yoke iron size based on the combination of virtual instrument and machine vision technology is designed to solve the problem of low accuracy of traditional manual detection. In the system, an image acquisition system based on message queue to realize the automatic acquisition of yoke iron images is developed by LabVIEW platform. In addition, image preprocessing techniques such as median filtering and binarization are used to filter image noise and interference, and then, Hough transform is used to extract yoke iron contour features, and the least square algorithm is used to measure yoke iron size with high accuracy. Finally, the designed system is used to measure the yoke iron size. The measurement results showed that the yoke iron size error is between 0.0001 mm and 0.0045 mm, and the repeatable measurement accuracy is high, which verified the effectiveness of the system.