The Research On Image Compression Coding Algorithm Based On Wavelet Transform

With the wide application of image acquisition device and the improvement ofacquisition resolution, image data are growing rapidly. In order to utilize the image dataeffectively, the compression of image and video has become an urgent problem and hasbecome a research hotpot in multimedia technology field. The wavelet transformtechnology becomes widely used in image compression fields for its goodtime-frequency partial characteristic and wavelet multi-resolution characteristicmatching well with the multichannel model of HVS. The image compression methodbased on wavelet transform has become an important branch of image compression，study and improve the algorithms of image compression based on wavelet is not only animportant task but also a research hot.A deep research on the EZW algorithm and some improvement are made in thispaper.First, this thesis introduces the research background and the signification of thethesis, and briefly describes the status of image compression at home and abroad.Then,this paper introduces the necessity and possibility of the image compression as well asthe basic method of image compression，the assessment criteria of image compression,and the international standard of static image compression.Then, a detailed description of the wavelet theory is made. By studying the wavelettransform for sub-bands characteristic, the advantage of wavelet coefficients is analyzedand the basic idea of wavelet transform and frame of wavelet transform for imagecompression is presented.After that, the research is focused on the image compression based on embeddedzero-tree wavelet. This paper introduces the coding algorithm in detail, and shows theEZW coding process and experimental result. According to the result, the advantage anddisadvantage of the classic EZW algorithm are analyzed.At last, two improved algorithms are put forward to reduce redundancy existing inthe encoding and distortion existing in the reconstructed image. One method is fixedlength run-length coding to optimize the output of EZW, reducing the correlation amongzero-tree roots for the purpose of further improve the compression ratio. The otheralgorithm is improved EZW based on HVS. The improved aspects are as follows: on theone hand, CSF mask can be weighted by wavelet coefficients of different frequency sun-bands, taking advantage of the nonlinear of the HVS to improve the quality ofcompressed image; on the other hand, redundancy occurred when all of insignificantcoefficients of one significant coefficient are insignificant, an improved algorithm withextended coding symbol is presented to reduce the redundancy to raise the imagecompression ratio. The experiment results show that, the two improved algorithms canachieve the expected purpose.