计算机工程与应用 ›› 2012, Vol. 48 ›› Issue (7): 191-193.

• 图形、图像、模式识别 • 上一篇    下一篇

砂轮地貌双目图像特征点提取和匹配研究

张小锋1,万 斌1,黄 娟2   

  1. 1.南昌航空大学 信息工程学院,南昌 330063
    2.浙江省气象台,杭州 310021
  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2012-03-01 发布日期:2012-03-01

Grinding wheel topography’s feature extraction and matching based on binocular image

ZHANG Xiaofeng1, WAN Bin1, HUANG Juan2   

  1. 1.School of Information Engineering, Nanchang Hangkong University, Nanchang 330063, China
    2.Zhejiang Meteorological Observatory, Hangzhou 310021, China
  • Received:1900-01-01 Revised:1900-01-01 Online:2012-03-01 Published:2012-03-01

摘要: 特征提取和匹配是双目视觉中的关键点和难点。对SIFT和Harris两种特征点提取算法应用于双目砂轮地貌图像的角点检测进行了研究,通过对SIFT特征采用欧式距离进行匹配,对Harris角点采用零均值归一化互相关算法(ZNCC)进行模板匹配,采用RANSAC算法剔除误匹配特征点对。实验结果表明,SIFT算法在双目砂轮地貌图像上应用较Harris算法效果理想。

关键词: 特征提取, 特征匹配, 双目视觉, 砂轮地貌, RANSAC算法

Abstract: Feature extraction and matching is a binocular vision difficulties and key points. In this paper, SIFT and Harris feature point extraction algorithm is applied to two kinds of grinding wheel topography binocular image corner detection has been studied, through the SIFT features are matched using Euclidean distance, on the Harris corner with Zero-mean Normalized Cross-Correlation algorithm (ZNCC) for template matching algorithm using RANSAC feature points excluding the mismatching pair. Experimental results show that, SIFT algorithm in the application of binocular image grinding wheel topography than Harris algorithm is effective.

Key words: feature extraction, feature matching, binocular vision, grinding wheel topography, Random Sample Consensus(RANSAC)