For detecting the corners of the mammary gland blood oxygen image effectively, using the improved SUSAN+CSS corner detector algorithm in the image. In this method, firstly, remove the noise of the image by the morphology open operation and median filtering; secondly, segment the image by the threshold based on the gray value of the image and transform to the structure chart image, then detect the corners with the improved SUSAN corner detector algorithm; finally, combine these corners and get the corners of the mammary gland image with the CSS. The results show that the method can detect the corners of the mammary gland image effectively, which can offer some available parameters for the diagnosis of breast tumor.