| This paper mainly studies the problem of edge detection of images corrupted by noises. An algorithm of edge detector based on statistics is presented. The algorithm firstly supposes that the area in the window is smooth or has only two values, and then estimates the value that is in the major in the window according to the local statistical property of the values in the window, and estimates the other value by the principle of statistics; lastly, transforms the values in the window to the new value to decide if a edge is existed. It is showed by the computer simulations that the algorithm has good ability of resisting noise and the computation is simple. |
