Abstract: In this paper, a new method for image smoothing--the modified neighborhood averaging method is presented. With this method, not only the noise can be smoothed efficiently, but also the blurred edge of an image be sharpened. It is simple in computation, requiring no a priori knowledge and predetermined parameters. In order to evaluate the performance of the proposed method quantitatively, a figure of merit for the test image generated by computer is defined and used as a measure of evaluation. The performance of the method is also compared with that of the gradient inverse weighted smoothing method and median filtering by experiments. Finally, processed results of real world images are given.