Abstract: Aiming at the existing problems of inefficiency, leak and error in the detection of rail surface defects, this article presented a method to detect the surface defects of rail based on HALCON software, and implement functions of recognizing and processing of defect images. The method was to acquire rail surface images by a simulative rail test platform, preprocess the original images, detect edge, extract features by using HALCON software, calculate and calibrate the relevant geometrical information of defects, get the position of different kinds of defects finally. The experiment showed that this method was with high measuring precision and efficiency, and could meet the practical application, provided a new way for on-line measurement for rail surface defects.