In order to solve such problems as low efficiency, low quality and instability of gear surface defect detection, we designed a detection system based on machine vision, sensor coupling. By multisensory coupling, and then CCD camera image collection of gear products, using VS2010 to cooperate with Halcon library for a series of analysis and processing of images. At last, the results are fed back to the control end, and the rejected device is removed to the collecting box. The system has successfully identified defective gear. The test results show that this system can identify and eliminate the defects gear quickly and efficiently. It has reached the requirement of gear product defect detection line automation and has a certain application value.