基于MATLAB的JPEG量化算法研究

Research of Quantization Algorithm in JPEG Based on MATLAB

以DCT变换为核心的JPEG图像压缩关键技术在于量化表的选取,量化表的选取决定了图像信息的损失程度和恢复图像质量。由于MATLAB语言具备高效、可视化及推理能力强等特点,为教学研究带来了极大的方便性,但目前还没有一套完整的在MATLAB环境下实现JPEG压缩全部流程的系统程序。文中以模块化方式编写了一套基于MATLAB的J PEG图像压缩和恢复所必需的函数构成完整的系统,提出了一种简单易行的可控制压缩比算法。通过大量实验数据分析和比较了几种效果较好且易于实现的修正建议量化表的算法,并重点就如何结合人眼的视觉特性和根据图像自身的特点寻找较好的量化算法等问题作初步的探讨。

The key technique of the compression of JPEG based on DCT is the selection of the quantization,and the quantization determines the graphics ’ loss after compression and the recover quality. MATLAB has brought a lot convenience for its efficient and powerful inference ability, but there hasn’t been a good working condition system environment about JPEG with the MATLAB language so far. So a group of model functions to compress images which are written in MATLAB language is provided in this work. A simple controllable compression algorithms is raised in this paper.Different algorithms about how to improve the quantization are discussed with a large number of experiments data, and the way to find a better quantization matrix based on the specific property of human ′s vision for each image is discussed.