A new iris pre-processing method was proposed, which breaks the traditional localizing principle. The process of noise detection was achieved by converting the spatial domain into the frequency domain. The detailed steps were as follows: Firstly, the light spot within the pupil was filled in the original image. And then the image was unfolded into a rectangle, the circle detection was substituted by the point and line detection in the rectangle image to find the inner and outer edge, and the normalization was also performed. Finally, eyelash and eyelid were detected and disturbed region was marked as well by taking advantage of the direction selectivity of 2D-Gabor filter in the normalized iris image. The results prove that this method not only solves the problem of pupil carry-over or the texture loss caused by the inner edge which is not an ideal circle, but also sharply reduces the location time and enhances the accuracy of iris recognition.