Dissertation > Excellent graduate degree dissertation topics show

Image Compression of Lifting Scheme for Combining the Integer Wavelet Transform with the Vector Quantization

Author: ZhangLiYing
Tutor: GuoShuXu
School: Jilin University
Course: Electronics and Communication Engineering
Keywords: Wavelet Analysis Vector quantization Integer Wavelet Transform Image Compression Coding
CLC: TN911.73
Type: Master's thesis
Year: 2004
Downloads: 175
Quote: 0
Read: Download Dissertation

Abstract

The main contents are: Lifting integer wavelet transform and vector quantization combined image compression method is accomplished by: 1, enlarge the image before the transform coefficients appropriate (the papers magnification) to reduce the mantissa truncated amplitude of the transform coefficients, the choice a DB9 / 7 wavelet filter, three integer wavelet transform; DPCM coding, transform low-frequency wavelet coefficients, high-frequency wavelet coefficients in accordance with the highest level of the coefficient address addressing re-queue mapping into a tree structure vector. Quantization section uses a local search method, by adjusting the partial search range to adjust the quality of the restored image to the coding bit rate and restore the object of the image quality. In modern communication, image transmission has become an important content, the size of the transfer amount of information is one of the important reasons affects the transmission speed. In order to improve the communication speed, a necessary means of image coding and compression techniques to reduce the amount of data transmitted. Image data compression has become the urgent needs of technological progress, precisely because of this demand, making the image compression (encoding) algorithms and techniques to become a very active area of ??research in the past 30 years, and has achieved great commercial success . A lot of image compression coding method, according to traditional methods with emerging category. Conventional image coding techniques: a pulse code modulated (PCM the Pulse_Code Modulation) quantization method (quantization), spatial and temporal sub-sampling, coding for (spatial and temporal subsampling coding), entropy coding (entropy coding), predictive-coded (Predictive Coding), transform encoding (transform coding), vector quantization (VQ, Vector Quantization), sub-band coding (SBC Subband Coding). The new image coding techniques: fractal coding (fractal coding) the model base coding (model__based, coding) wavelet coding (wavelet coding). Various encoding method, has its own unique, has its drawbacks, if we have the traditional encoding method with the new encoding method organic combination, you will get a better quality of image compression. Spatial distribution characteristics of wavelet image coefficients and its high-resolution characteristics of wavelet image coefficients in the spatial location and content, which is very suitable for the use of vector quantization (VQ) techniques to deal with. VQ techniques can not only avoid the coding redundancy due to the the loose spatial distribution of wavelet image coefficients structure and complexity, and can better characterize the correlation between this coefficient, while the wavelet transform combined with vector quantization techniques can be overcome vector quantization coding three problems (1) it is difficult to generate a generic codebook; (2) LBG algorithm is a smoothing effect on the high-frequency component, thus reducing the resolution; (3) it is difficult to combine the characteristics of the human visual system. lt; WP = 46 GT; This thesis is image compression coding scheme Lifting integer wavelet transform combined with VQ image compression. IWT is a great success in the lossless coding and compression, lossy compression efficiency is much lower than the traditional DWT. Especially IWT is based on the the rational transformation parameters (Rational Parameter) in the SPIHT or SPECK encoded framework under lossy compression peak signal-to-noise ratio (PSNR) than a difference of 3-6dB a DB9 / 7 DWT One of the reasons is: truncated rounding the (mantissa rounding) makes the IWT become non-linear transformation (Nonlinear Transform), reducing the energy concentration of the IWT. The second reason: no transformation parameters for rational IWT normalization process, resulting in the transformed image coefficient does not like the coefficients in the DWT decomposition directly on behalf of each sub-band energy. Based on the above issues, this paper first transform ago image coefficient appropriate to enlarge the mantissa truncated to reduce the amplitude of the transform coefficients; transformation parameters of rational numbers the IWT the introduction of a scaling factor to improve the energy concentration of the IWT; Finally, adjust the zoom factor, the IWT lossy compression effect best. Select a 512 pixel x 512 pixel standard test chart (Lena) magnification, the pixel values, and then three integer wavelet transform, the choice a DB9 / 7 wavelet filter. The low-frequency wavelet transform coefficient DPCM coding, transform high-frequency wavelet coefficients for addressing re-queuing mapped into a tree structure vector in accordance with the highest level of the coefficient address. Static image of a representative selection of 10 as a training sequence using the LBG algorithm generated the vector codebook, and the codebook to be sorted in accordance with the size of the root node of the vector formed orderly codebook. Quantify using a local search method. This method can reduce the computational complexity After simulation of judgment, to achieve fast wavelet transform, the use of local search method to quantify, 8-12 times than the traditional global search method to improve the speed, and you can adjust the local search range to adjust the quality of the restored image, to achieve the encoding speed and restore the object of the image quality. As can be seen from the simulation results, in the case of compression is relatively small, the restored image of the JPEG standard algorithm having a higher PSNR, SNR, but the subjective observations, the present algorithm to recover the same image with the restored image with the JPEG algorithm subjective quality; compression ratio, the recovery of the JPEG standard algorithm image box effect. Subjective visual experience is acceptable for the algorithm to restore the image. Thus, the study in this paper the method is feasible.

Related Dissertations

CLC: > Industrial Technology > Radio electronics, telecommunications technology > Communicate > Communication theory > Signal processing > Image signal processing
© 2012 www.DissertationTopic.Net Mobile