Contourlet Based Lossy Image Coder with Edge Preserving

In this paper we present a lossy image coder based in multiscale and directional contourlet transform which allows edge preserving in decompressed images. The preserving task is made by capture the edge points of an image using a robust and efficient edge detector known as Smallest Univalue Segment Assimilating Nucleus (SUSAN). The resulting edges coordinates are used as a priori knowledge of the positions of points which are mapped to the correspondent position in the transformed domain by exploit the contourlet parent-child relationship. The contourlet points are coded and finally the image is decompressed to measure the quality. The use of contourlet transform allows us to take advantage of the directional and multiscale representation of edges which is showed in the good edge preserving results obtained. Key-Words: Image compression, edge preserving, transform domain, contourlet, parent-child relation.