Iris location is a very critical step in iris recognition. In this paper, we chose a primitive model to locate the inner edge of the iris. In order to make the extracted inner edge more accurate, the iris image of the collected image was divided into light and normalized, then the window function and the double threshold were set to define the primitive model. Edge texture information for edge location. Finally, the iris images of CASIAIris, UBIRISV1 and MMUIrisV1 iris database were selected and tested on the MATLAB2010b with Daugman and Wides. The experiment proved that this method was very accurate and timeconsuming. The precise positioning of the inner edge of the iris has a good application value.