| With the strong demand for multimedia data communication, image coders should exhibit excellent distortion-rate (D-R) performance while at the same time providing desirable functionalities for a variety of applications. Subband coding, based on the wavelet theory, provides a multi-resolution decomposition of images, and it has become a concrete foundation of state-of-the-art image coding techniques. A class of image coders in this wavelet domain, consisting of the embedded zerotree wavelet (EZW) coders, has shown excellent D-R performance with low computational complexity, while generating an embedded bit stream. Likewise, tree-based fractal coders are very efficient in representing the parts of an image that have self-similarity. In this work, we propose two hybrid fractal zerotree wavelet image coding algorithms. The algorithms couple an EZW coder and a fractal coder in the wavelet domain. In addition to improving compression performance, the proposed algorithms also allow one to impose desirable properties from each type of image coder.; Many new applications require the transmission of imagery over relatively hostile channels such as the Internet or wireless mobile networks; error resilient image coders that also provide good compression performance are strongly required. However, most state-of-the-art image coders are very sensitive to errors and modifications must be made for use with a noisy channel. A robust image coding algorithm that can be applied to any zerotree-based coder is proposed. With this algorithm, a very explicit segmentation and packetization method for an image bit stream is developed. Lastly, a novel robust fractal zerotree wavelet image coding algorithm that can recover lost data is developed. This is, to the best of the author's knowledge, the first use of fractal coding as a type of forward error correction. |
