J4 ›› 2012, Vol. 50 ›› Issue (01): 95-100.

• 计算机 • 上一篇    下一篇

双目立体视觉匹配的预处理技术

常淑华1, 宋艳秋2, 杨永敏3   

  1. 1. 吉林农业科技学院 信息工程学院, 吉林 吉林 132101; |2. 长春市希望高中, 长春 130022|3. 哈尔滨工业大学 机器人技术与系统国家重点实验室, 哈尔滨 150080
  • 收稿日期:2011-06-27 出版日期:2012-01-26 发布日期:2012-03-06
  • 通讯作者: 杨永敏 E-mail:hitymyang@163.com

Matching Preprocessing Technology of Binocular Stereo Vision

CHANG Shuhua1, SONG Yanqiu2, YANG Yongmin3   

  1. 1. Department of Information Engineering, Jilin Agricultural Science and Technology College, Jilin 132101,Jilin Province, China|2. Changchun Hope |High School, Changchun 130022, China;3. State Key Laboratory of Robotic and System, Harbin Institute of Technology, Harbin 150080, China
  • Received:2011-06-27 Online:2012-01-26 Published:2012-03-06
  • Contact: YANG Yongmin E-mail:hitymyang@163.com

摘要:

针对双目立体视觉图对区域匹配应用中存在的问题, 提出一种三阶段的匹配预处理方法. 该方法将小波阈值去噪算法和直方图均衡化技术相结合, 有效地去除了图像随机噪声和高斯噪声, 实现了立体图对间亮度差异的平衡, 并经过拉普拉斯锐化处理提高了图像对的对比度. 实验结果表明, 三阶段匹配预处理算法可有效提高立体图对间的匹配准确率.

关键词: 双目立体视觉, 小波阈值去噪, 直方图均衡化, 拉普拉斯算子

Abstract:

To solve the problem existing in the application of binocular stereo vision imagepair, a threestages matching preprocessing method was proposed. This method combines the wavelet denosing algorithm and histogram equalization method effectively. The random noise and Gauss noise of image are filtered and luminance difference between stereo imagepairs is balanced at the same time. Finally the contrast of imagepairs was improved by Laplace sharpen processing. The effectiveness of threestages matching preprocessing algorithm was proved by experiment. The application of the algorithm could improve the matching accuracy of stereo imagepairs effectively.

Key words: binocular stereo vision, wavelet threshold de\, noising, histogram equalization, Laplace operator

中图分类号: 

  • TP391.4