Abstract: There is no strict mathematic definition for image noise till now, general noise filtering methods have the character of blindness. This article defines that the area around the point p but excluding the point p is the background (the surrounding environment) of the point p. It maintains that the noise point is not consistent with the background change rules. Thus, this paper proposes a method for filtering based on the background change rules. The main idea is to use the change rules around the noise point to “rule” the noise point, making it become harmony with the surrounding environment, so as to eliminate the noise. This method uses the surrounding environment to correct the noise point, letting the noise in accordance with the surrounding environment, thus not damaging it. The edges of the image is also the environment, therefore the method is a lossless filtering for the edges of the image, so it can protect the information of the edges of the image.

Key words: image denoising, background change rules, filtering, the edge of the image