Morphological Zerotree Compression Coding Based On Integer Wavelet Transform For Iris And Fingerprint Image

The iris recognition and the fingerprint recognition technologies have some fine characteristics, such as universality, uniqueness, permanence, living body, antifalsification, so they almost become the best biological recognition technologies, and they become the key points which the domestic and foreign companies and the scientific research units study. As the iris recognition and fingerprint technology enter the commercial application domain progressively, the compression coding of the iris and the fingerprint images has been put forward as a new request in order to complete some actual needs, such as long-distance image transmission, image database storage.The iris and fingerprint image belong to the texture image. The texture growth characteristic and the irregularity which the iris and fingerprint texture present will cause the iris and the fingerprint image compression code to be different from the general image compression coding method. Therefore which kind of compression algorithm is adopted will be effective on the texture image and whether the high compression rate and the fine recovery result will be guaranteed will be a new topic of the iris recognition and the fingerprint recognition domain.In recent years, the wavelet transformation with its reducing correlation ability and outstanding time frequency partial ability has obtained the unprecedented success in the image compression coding domain. Embedded zerotree wavelet code (EZW) is one of the quite famous wavelet encoders, which was proposed by J.M.Shapiro in 1993. At present, it is considered as one of the most advanced image compression code methods. The advantage of the EZW method mainly depends on utilizing the direction zerotree structure effectively, causing the massive trees structure zero coefficients only to use a zerotree root to indicate, which has obtained the high compression coding efficiency. But many "isolated zero" that could not contain in the zerotree have to carry on the complicated expression.In 1999, Sergio D Servetto, etc. proposed an algorithm——MRWD (Morphological Representation of Wavelet Data), which is different from zerotree ideological system. Comparing with EZW, MRWD effectively makes use of the characteristic that signal energy distribution is centralized within wavelet territory, namely the important coefficients assemble in each subband. MDWD with the morphological dilating operation, by means of using the structural element to control the shape and size of the clusters, directly establishes the non-regular cluster structure of the important coefficient in each subband , avoiding the complicated expression of "isolation zero" in EZW, which obtains the result superior to EZW. Through comparing the above two famous wavelet image compression code algorithms, we can find out that they only stress on one aspect of the wavelet coefficient data construction and the statistical property, but in fact, the wavelet coefficient data constructions and the statistical properties are too many, such as the important wavelet coefficient clustering inside the frequency band, the important wavelet coefficient similarity between different frequency band, the value weaken characteristic between bit plane frequency band, the important wavelet coefficient similarity in the same frequency band of different bit planes. This article utilizes synthetically each kind of statistical properties of the wavelet coefficient, realizing the morphology zerotree compression code which is based on the bit plane decomposition. If this algorithm has too many bit planes the decomposition will cause the complexity of the algorithm and too long running time, so we propose the integer wavelet transform to realize the decomposition. This thesis proposes the morphological zerotree compression coding based on integer wavelet transform. It obtains the less bit planes, and the integer wavelet coefficient matrix that does not need the quantification. At the same time, the integer wavelet transformation is reversible, therefore, and it will realize...