In order to overcome the shortcomings of traditional noise removal methods which the remaining noise is still large, an algorithm which was based on the local feature of the image was introduced for removing the Gaussian noise according to the impact of Gaussian noise on visual images. Firstly, the pixel was or wasn’t lie smoothing domain was estimated based on the local pixels’ continuity in the image. Secondly, the edge and texture of the image were extract by morphologic according to the local continuity properties of texture and edge, then the pixel contaminated with Gaussian noises was estimated. Lastly, the noise point in the smoothing domain was removed making use of the adaptive neighborhood, local smoothing for the edge or texture pixel. Comparing to the traditional method, this algorithm could preferably remove the noise and improved image visual effect.
