| As the method of passing information, image information is very important. However, the predominant drawback to apply digital image is the bulk data that need many bits. In order to overcome that drawback, we must compress the digital image. The definition of image compression is the processing of decreasing original data in the position of insuring the image quality. Because of the redundancy of digital image data, we can compress the image. Wavelet analysis have a variety of advantages, so there are a lot of image compression methods based on wavelet transform.The basic aim of image compression is minimizing the distortion with the needed Bit-Rate. The image compression based on wavelet transform is a valid method, but because it uses subbands' redundancy, the complexity is so much that the compressing process consumes much time. So the efficiency of wavelet coding depends on bit allocation, that is , definite bits must be allocated to a serial of quantizations validly. By using kinds of quantization tables, each quantization can work at kinds of rates. In a word, bit allocation is the method deciding how to allocate definite bits to a variety of subbands. A excellent bit allocation can minimizing the overall distortion..The emphasis of this thesis is studying the methods of bit allocations of digital image compression based on wavelet transform and managing to minimizing the overall distortion. We know from the Rate-Distortion theory that the aim is gainable. In the thesis, I describe the theories of two bit allocation methods and also give the realizable algorithms. The difference of the two methods is that the first is in the position of convex Rate-Distortion and the second is in the position of arbitrary Rate-Distortion curve. The results is gave in the character five, and in the character, I compare results of the thesis with those of JPEG, JPEG2000, ZTE and also conclude from the comparison of results that get form the two bit allocation.
|
PDF Full Text Request