Abstract: Detecting points of interest is one important issues in image processing systems and these points could be uses to eliminates the information of the images to minimum number to be used in recognition or data reduction ..etc. in this paper we produce a method to detect corners in images depending on topological and tint information , The gradient of intensity is calculated in two steps one for grayscale to detect curvatures and other for binary to detect corners in curvatures . The method was quite accurate , efficiency and fast , but rather fair with noisy images