In order to improve the efficiency of piston rod surface flaw detection, a fullautomatic piston rod surface flaw detection system based on machine vision was designed. The Vision module of LabVIEW software was used for image acquisition, processing and recognition, and the TIA portal software was used for motion control. The binarization algorithm was used to filter out minor defects. The image was corroded and then expanded to remove isolated spots, burrs and bridges. The experimental results show that the system is stable, reliable and efficient