In order to solve the problem of low accuracy and great lag in the inspection of hollow filter bar end face, a vision inspection system for hollow filter bar end face was studied based on machine vision technology. Using PLC to read the pulse signal on the encoder in real time, the industrial array camera was triggered to collect the image of the end face of the hollow filter bar. Image optimization technology was used to process the image of the filter bar end face to ensure the image quality. The dimension and roundness of the end face of the hollow filter rod were measured by using the least square method. Through the method of template matching difference, whether there are defects in the end face of the hollow filter rod is detected, and the alarm is displayed on the man-machine interface. The test results show that the system can detect the dimensions, roundness and defects of the end face of the hollow filter rod in real time, and the detection accuracy can reach 99% at different speeds, which lays the foundation for automatic visual inspection of the filter rod and improves the qualified rate of the product.