Digital image coding with high compression ratio

Presents an efficient method for coding digital images with high compressions and acceptable reconstructed images. The information of an image is divided into two main components, namely, the low-frequency component containing the image textured regions and the high-frequency component containing the image features or edges. The coding strategy is then based on the feature-texture description of the image. Low-pass filtering is applied to obtain the low-frequency component which is coded by conventional transform coding. To obtain the image features, the authors propose an algorithm based on the local statistical of the image in conjunction with the Gaussian smoothed Laplacian operator for edge detection. They then decompose the edge information into 4 directional segments. Thresholding is performed to reduce redundancy before using run-length coding for each directional segment. Computer simulations show that the proposed method is simple to implement and possesses satisfactory performance