The Technology And Implementation Of Hybrid Image Encode

With the popularization of the Internet and the improvement of the image application and the great capacity of the digital image and video data, people are always zealous of the research of the digital coding technology. After decades years the first era coding technology based on the three classical coding algorithms have reached it's terminal. The second coding technology based on the HVS (Human Visual System) and the physiology can lead us to a new areas. The wavelet transform is the milestone of the second coding technology. It can be applied into image texture analysis, image coding, computer vision, speech recognizing, earthquake signal processing and so on. S.G.Mallet introduced the multi resolution analysis into the computer vision domain.. In 1986, he summarizes all wavelet 's construct method, and give a discrete wavelet transforms and construct algorithm based on the band filter.The signal is decomposed with different layers by wavelet decomposed in the space domain and each layer has different resolution. Each layer's relativity is depended on the data self absolutely after decomposed, because of the orthogonal of wavelet. So the relativity of inner method decomposed is excepted. The decomposed of wavelet transaction in the space domain with many layers ,at the same time the frequency domain is decomposed. At the frequency domain of each layer, distributing of high frequency and low frequency is concerned with the direction of the frequency distributed in the original data. The energy can be disparted different fields according to the frequency after the data is decomposed by wavelet. Here the low frequency expresses the main information in the original image and it has an important effect on the image restoreing. The distortion of the low frequency part will result in the distortion of the whole image reconstruction and affect the image quality badly. The strategy of this paper is to code the low frequency part by forecasting-difference method and can get compressing without loss. The specific information is focused on the high weight. The wavelet parameters in same space situation of different resolutions show the specific information of the same object and has the self-correlativity. It is proper for the embedded code. The Embedded Zero-tree wavelet code is to judge isolated parameter and zero-tree and to code entropy. The judging threshold value will descend layer by half layer upon layer.In EZW we introduce the concept of symbol. It can express the character of every wavelet parameter. Zero-Tree Root(ZTR):In this quantization, the range value of this wavelet parameter is smaller than a given threshold T and the range value of its offspring is smaller than the given threshold T. At the same time, the parent node coefficient is larger than the range value of this quantization.POS(positive): In this quantization, the range value of this wavelet parameter is larger than a given threshold T and its sign is 1.NEG(negative):In this quantization, the range value of this wavelet parameter is larger than a given threshold T and its sign is 0.Isolated Zero (IZ):In this quantization, the range value of this wavelet parameter is smaller than a given threshold T but existing offspring whose value is larger than the given threshold at any rate . At the same time, the parent node coefficient can be any value.The image which has been transacted with wavelet has formed a tree structure according to the frequency from low to high. The low frequency band will be coded as DPCM and high frequency band will be coded as EZW.Two separate lists, the dominant list and the subordinate list, are maintained both at the encoder and the decoder sides. At any point of the process, the dominant list contains the coordinates of those coefficients that have not yet been found to be significant. While the subordinate list contains magnitudes of those coefficients that have been found to be significant. The process that updates the dominate list is called the significant pass, and the process that updates th...