| 注册
首页|期刊导航|计算机技术与发展|共生矩阵纹理特征的DSP优化实现

共生矩阵纹理特征的DSP优化实现

王职军 梁光明 刘任任 徐克强 谢俊

计算机技术与发展Issue(11):13-16,22,5.
计算机技术与发展Issue(11):13-16,22,5.DOI:10.3969/j.issn.1673-629X.2014.11.004

共生矩阵纹理特征的DSP优化实现

DSP Optimized Implementation of Co-occurrence Matrix Texture Feature

王职军 1梁光明 2刘任任 1徐克强 1谢俊1

作者信息

  • 1. 湘潭大学 信息工程学院,湖南 湘潭 411105
  • 2. 国防科技大学 电子科学与工程学院,湖南 长沙 410075
  • 折叠

摘要

Abstract

Texture features based on Gray Level Co-occurrence Matrix ( GLCM) has the drawbacks of complex calculations and time-consuming,which seriously affects the efficiency of program execution. Aiming at this problem,analyze the principle of the extracted tex-ture features based on GLCM in this paper,study on the performance of the TMS320C6678 DSP,propose the optimization method of ac-cess bandwidth and software pipelining based on GLCM texture feature,then achieve the optimization of GLCM-based texture features by using of TMS320C6678 in the CCS5. 3 software platform,so that the program execution time is reduced from 1. 94 ms to 0. 259 ms. Ex-periment results show that the proposed optimization method can shrink code execution time, improve code performance and meet the needs of embedded image processing system.

关键词

共生矩阵/纹理特征/TMS320C6678/优化

Key words

co-occurrence matrices/texture features/TMS320C6678/optimization

分类

信息技术与安全科学

引用本文复制引用

王职军,梁光明,刘任任,徐克强,谢俊..共生矩阵纹理特征的DSP优化实现[J].计算机技术与发展,2014,(11):13-16,22,5.

基金项目

湖南省自然科学基金(07JJ6115) (07JJ6115)

计算机技术与发展

OACSTPCD

1673-629X

访问量0
|
下载量0
段落导航相关论文