It is important to detect defects of surface of railway train bearing roller. It also has relationship with transport security. The surface defect detection system of train bearing rolle has been described. This paper states the working principle and structure of the magnetic particle testing system based on the image processing technology. This paper has discussed the image of image processing steps and algorithms for magnetic defects picture. Image processing is the core of the system. It consists of image pre-processing, image segmentation, feature pick-up, defects recognition. It also detailed introduces the experimental verification, and analysis the measurement error from the images perspective. The results show that this method can effectively detect common defects on the surface of the roller.The detection system has many advantages as follows: high resolution, high speed, simple hardware, flexible and abundant functions of software, strong anti-interference. Meanwhile, it is outstanding characteristics, such as combinability and expandability. It is of great practical importance, as to enhance the safety between the process of testing , improvement of detection level, measurement accuracy and productivity, great reduction of people's labor intensity. |