Image Coding System Design Based On Vector Quantization

As an effective technology for data compression, vector quantization (VQ) is widely used in the field of image coding because of its simple algorithm and high compression rate, especially ASIC chip for image coding. The algorithm of VQ must be improved to reducing amount of calculation, increasing compression rate and steadying the quality of rebuilt image at same time. In order to meet the requirement of image transmission in real time, it is necessary to design this image coding system in ASIC method.According to analyzing three basic technique (size of image block, codebook and match algorithm) of VQ, this paper provides three novel algorithms improving coding performance. This paper also integrates three algorithms into a complete VQ image coding system. At the end to this paper, an ASTC circuits scheme of this system is put forward.Firstly, based on conventional VQ algorithm, a novel algorithm called partition dynamically VQ (PDVQ) is presented, which can not only can reduce amount of calculation and increase compression rate, but also keep stable PSNR values.Secondly, according to integrating PDVQ with classified image coding algorithm and equal_sum_block match algorithm, a complete program of VQimage coding system is brought forward. This program combines all advantages of three algorithms and meets the requirement of image transmission in real t ime.Lastly, based on ASIC design flow, the RTL level circuit scheme of the VQ image coding system is suggested and synthesized in the available technology library.Simulation and test result of VQ image coding system shows that coding speed is increased to 3 to 6 times. For images with different details, the compression rate is increased to 10% to 100% respectively. And all the improvement is achieved in the presupposition of steady PSNR of rebuilt image.