Abstract: The automatic defect detection technology for resin lenses was studied, including its hardware system design and its image processing method. The image difference method was used to remove the edge of the lens image in order to obtain high detecting speed, the effect of dust in lens detection was solved, and an improved normalization method was also proposed to determine the number of defects, which could provide technical support and feasibility for the development of subsequent digital control module of the classifying and sorting system of the whole resin lenses defect detection system. Detection accuracy of the detection system is 0.03 mm, speed is 1 piece/s, and the measuring error is maintained below 5%.