Abstract: This paper proposes an improved SUSAN algorithm to improve the tungsten ore processing rate and qualification rate in the preliminary tungsten process. Based on SUSAN algorithm, the method of selecting adaptive threshold t and rapid extracting is put forward by using unique color and textural characteristics of wolframite image.The method is realized by VC ++ programming which is finally applied to edge detection extraction of tungsten ore image. Experiments show that image edge detection has advantages of high accuracy and fast operation. In comparison with the previous method, this method improves the calculation speed by improving the detection results. It has practical significance for the preliminary tungsten processing.