Abstract:There are some problems such as slow speed, high rate of defect omission and false detection when the existing manual visual inspection method is used to inspect the surface quality of precisionmachined carbide micronozzle. In order to solve the above problems, a method for detecting nozzle image defects based on machine vision was proposed. The nozzle defect image type and nozzle structure were analyzed, the nozzle defect circle edge fitting, scar edge enhancement, polar transformation and defect gray value difference statistics were heavily studied. By using this method, a lot of calculations for defect location and detection caused by the complex nozzle structure were avoided. Through the detection of qualified and unqualified nozzles, the accuracy of the defect detection method was verified to be 98.6% and the detection time of each piece was about 0.834 s, while the accuracy of the manual visual inspection was about 91.2%, and the detection time of each piece was about 5.213 s. So this algorithm can effectively improve the detection accuracy and speed, and meets the requirements of nozzle detection accuracy and realtime performance in industrial production lines.