An image compression scheme using wavelets-based edge extraction and low-frequency component expansion

A simplified Mallat (1989) algorithm is proposed to extract the edge information for image coding. An input image is first separated into a smoothed image that comprises low-frequency components only and an image for strong edge information. The low-frequency image and the edge one are then encoded by using the DCT and run-length algorithms, respectively. By using this strategy, the artifacts such as edge blur and block effects that are usually present in traditional DCT scheme especially at low bit rate, are overcome. It is shown that the proposed coding scheme not only gives a reconstructed image of high quality but enjoys a low computational complexity and a simple realization structure as well. Therefore, it is expected to be implemented easily with hardware