吉林大学学报(信息科学版) ›› 2015, Vol. 33 ›› Issue (5): 510-.

• 论文 • 上一篇    下一篇

基于遗传算法的优化DCT 图像压缩方法

刘媛媛1,2, 陈贺新1, 赵岩1, 孙红岩3   

  1. 1. 吉林大学通信新技术重点实验室, 长春130012;2. 吉林农业大学信息技术学院, 长春130118; 3. 长春理工大学电子信息工程学院, 长春130022
  • 收稿日期:2014-12-01 出版日期:2015-09-30 发布日期:2015-12-30
  • 作者简介:刘媛媛(1980—), 女, 长春人, 吉林农业大学讲师, 吉林大学博士研究生, 主要从事图像和视频处理、视频编码研究, (Tel)86-13504456332(E-mail)liuyuanyuan1980@126. com; 陈贺新(1949—), 男, 吉林大安人, 吉林大学教授, 博士生导师, 主要从事图像和视频编码、多维矩阵理论与应用研究, (Tel)86-13086825533(E-mail)chx@ jlu. edu. cn。
  • 基金资助:

    国家自然科学基金资助项目(61171078; 61271315)

DCT Optimization Image Compression Method Based on Genetic Algorithm

LIU Yuanyuan1,2, CHEN Hexin1, ZHAO Yan1, SUN Hongyan3   

  1. 1. New Communication Technology Key Laboratory, Jilin University, Changchun 130012, China;
    2. College of Information Technology, Jilin Agricultural University, Changchun 130118, China;
    3. School of Electronic and Information Engineering, Changchun University of Science and Technology, Changchun 130022, China
  • Received:2014-12-01 Online:2015-09-30 Published:2015-12-30

摘要:

为打破传统DCT(Discrete Cosine Transform)变换矩阵统一的束缚, 从离散余弦变换DCT 的基本原理及特点出发, 针对图像信息对DCT 变换矩阵进行优化处理, 实现对图像的高效压缩。将图像进行DCT 压缩后重构并与原图像进行比较; 利用遗传算法求最优解使其均方误差最小以优化DCT 变换矩阵的系数; 以优化后的变换核对图像进行处理。实验结果表明, 对线性边缘、纹理特征图像小分块处理基于遗传优化DCT 的算法在种群个数为40 ~ 60、字符串长度为8、交叉概率为0. 7 ~ 0. 8、变异概率为0. 007 ~ 0. 008 时, 图像压缩效果达到最佳。

关键词: 优化DCT, 遗传算法, 图像压缩, 均方误差

Abstract:

This paper is based on the DCT(Discrete Cosine Transform) basic principle and characteristics, andits purpose is to break the shackles of the unified characteristics of traditional DCT transform matrix, for theoptimization on the DCT transformation matrix of image information, to achieve the image compression efficiently.Firstly, compare the reconstructed image by DCT with the original image. Then, use genetic algorithm to find theoptimal solution by minimizing the mean square error, in order to optimize the DCT transform matrix coefficient.Finally, use the transform kernel optimized for image processing. Experiment was carried out from blocking theimage and setting genetic initial parameters, and the experimental results indicate that for the linear edge andtexture feature images of small block processing, when the number of population is 40 ~ 60, the length of thestring is 8, the crossover probability is 0. 7 ~ 0. 8, and the mutation probability is 0. 007 ~ 0. 008,the DCToptimization method based on genetic algorithm for image compression to reach the best effects.

Key words: discrete cosine transform ( DCT) optimization, genetic algorithm, image compression, mean
square error

中图分类号: 

  • TN919. 85