Study Of Image And Video Coding Based On Wavelet Transform

The image and video is the most important medium of information. However the prominent character of them is with the huge data, as brings on great difficulties to the storage and transmission of image data. With the fast development of remote sensing technology, the image data increased rapidly. The contradiction between the limited channel capacity and the huge data of remote sensing image is more and more prominent. As an effective method, image data compression is paid great attention to by the researchers in the field of remote sensing.The analysis of wavelet is developing in the 1980s to deal with unbalanced signals. On the basis of multi-resolution, it overcome the default of Fourier analysis. The study of wavelet-based image and video compression has drawn much attention in this field. The image and video coding based on wavelet, transform is studied in this thesis. The new algorithm of wavelet-based image and video compressing is proposed and tested. The study content and key contribution of the dissertation is as follows:1. Introducing the theory of wavelet transform, the concept of multi-resolution analysis, and the Mallat algorithm. The signal of Nossin is composed and decomposed by Mallat algorithm.2. The advance of wavelet transform is further analyzed. Wavelet transform can change the distribution of image energy and make the energy concentrating. So It can avoid the diamond effect and utilize the feature of progressive transmission. Lena image is decomposed and reconstructed by DWT(Mallat algorithm in 2-dimension, CDF9/7 wavelet)3. KZW, SPIHT and SPECK is illustrated, and the flow chart is given in this thesis. Standard Couple image is decomposed and reconstructed with EZW and SPIHT at low bit rates. It is tested that the SPIHT is the best at PSNR, and the subjective sense.So, the SPIHT is the best at low bit rates.4. The pattern is proposed to mend SPIHT from three aspects: spatial orientation tree, Human Visual System, and the composing of SPIHT andSPECK. The flow chart of the mended SPIHT is proposed and the fast lifted wavelet transform is used. Lena image is decomposed and reconstructed by SPIHT and the mended SPIHT at low bit rates below 0.1 bpp. And the mended SPIHT is compared to the SPIHT with the increasing of bit rates. It is concluded that mended SPIHT is better than SPIHT at low bit rates, otherwise the latter is better when the bit rates is above certain value.5. A new algorithm of video coding based on wavelet transform. The compression pattern based on DWT and DCT is proposed and the method of three step search is improved. Finally, the validity of this algorithm is tested by examinations.