Research On Recognization Of Roof Eyewinker Based On Image Process

With the rapid development of world’s high-speed railway, in order to conform to developing trend of national economy and meet people’s needs China railway also gets fast growth. In recent years, with China’s railway operation speed greatly enhance, the safety of the locomotive running attain more and more attention by the industry. The roof eyewinker-image of locomotive are very important as a part of locomotive on-line detection. This thesis put forward eyewinker-detection system based on image of locomotive roof to identify the foreign body and the damage of equipment on the roof. The thesis introduce the roof eyewinker detection and recognition principle combined with the actual need to set up the image acquisition system of the locomotive roof, and then detailed algorithm of the image processing is given.In image processing stage each module of the proper algorithm used for locomotive roof image processing are proposed. The main algorithms involved are as follows:First of all, light compensation the original roof image is done. In order to eliminate the global or local uneven illumination of the roof image, a grayscale histogram equalization and homomorphic filter method are used.Secondly, phase correlation method based on polar-log coordinate is used to complete registration of the roof image. The method can ensure accuracy and efficiency of the detection of translation, rotation and zoom between the roof images and the template.Thirdly, foreign body recognition is done through the contrast between the tested image and standard template image. Iteration threshold of binary image processing is used to eliminate interference.In the end, every foreign body is assigned a unique labeling value. This step proposes two improved algorithm based on4-neighborhood connected component labeling algorithm with the combination of high accuracy and efficiency.The thesis use matlab to simulate algorithm and finally set up image processing platform with MFC. In the end of the thesis, the detection system is summarized and the prospect of the research is given.