Nowadays, the installaton of bake for railway locomotive is very complicated. There are many parts around the brake. It is very difficult to locate the brakes position and to detect its wearing situation automatically. Aimed to solve the problems, a new method for the bake detection of image based on the DNA algorithm has been proposed. Firstly, the experimental system for brake images capture was setup. When the locomotive ran at a certain position, the system could capture images automatically and save them to the hard disk in computer. On this basis, the median filtering and gray enhancing operations were done to the images. A segmentation mehod based on the Deoxyribonucleic Acid (DNA) algorithm has been researched and applied to the brake images of threshold value. Compared with the traditional ostu method of threshold value, the new method could locate the brake correctly, acquire its regional threshold value effectively and segment the brake infomation. Then the math morphologic processing was done to the segmented images and the isolated points around the brake could be filtered. In the end, the edge dection method based on the sobel algorithm has been operated, and the clear ideal edge of brake is extracted, which make some basis for automatic detection of brake for ralway locomotive.