The use of low-level visual features to search and retrieve information in the multimedia databases has drawn much attention in the recent years [1,2]. Many of the existing techniques of image retrieval are based on image segmentation, which is a difficult task in many practical situations due to image noise and various compression artifacts. In this paper a novel approach to the problem of edge preserving smoothing, which allows to break an image into a set of homogeneous regions, is proposed and evaluated. The new algorithm is based on the combined forward and backward anisotropic diffusion with incorporated time dependent cooling process. This method is able to efficiently remove image noise while preserving and enhancing image edges. The proposed algorithm can be used as a first step of different techniques, which are based on color, shape and spatial location information, to search and retrieve information from multimedia databases.