Edge preserving filters - Aid to reliable image segmentation

Image segmentation is the key to image understanding/scene analysis. An effective image segmentation can be best accomplished by maximizing the differences between unlike regions while simultaneously minimizing the variations within each of these regions. The former can be achieved by enhancement and detection of boundaries between segments and the latter by intrasegment smoothing. The traditional smoothing techniques, while attaining the intrasegment smoothing, tend to blur the intersegment boundaries, thereby diluting the effectiveness of the image segmentation process. The reported investigation is concerned with the development of an integrated edge preserved smoothing procedure which reduces the risk of delineation of noise-triggered edges. Aspects of methodology development and implementation experience are discussed.