一种改进的车头SURF特征提取和匹配  

Car Feature Extraction and Matching in the Region of Car Face Based on Improved SURF

作　　者：刘丹丹[1] 魏维[1] 

机构地区：[1]成都信息工程学院计算机工程学院,四川成都610225

出　　处：《成都信息工程学院学报》2014年第2期167-172,共6页Journal of Chengdu University of Information Technology

基　　金：四川省教育厅重点科研资助项目(09ZA155)

摘　　要：通过对加速鲁棒性算法特征关键点提取和匹配的研究,提出一种改进的车头加速鲁棒性特征点提取方法,该方法在车头图像像素梯度变化较大的范围内提取关键点,然后采用最近邻比例算法进行特征点匹配.实验结果表明,提出的改进方法不仅能够筛选出鲁棒性较强的关键点,同时提高了特征的提取速度和匹配效率,在总时间上比尺度不变特征变换算法提高了3～6倍;比加速鲁棒性算法提高了近1/4.Through the study of speed up robust feature（SURF） extracting and matching key points,a new method of extracting car key points based on SURF has been proposed in this paper.This method extracts key points that are obvious different from the surrounding pixel gradient in the region of car face,then matches the feature detectors by using K-Nearest Neighbor ratio.Experimental results show that,the method proposed in this paper can select the robustness key points,at the same time,improve the speed of feature extraction and matching.The total time by using the method proposed in this paper is increased 3-6 times contrast to Scale Invariant Feature Transform,and is approximately one fourth of speed up robust feature.

关 键 词：计算机应用技术 图形图像处理 特征描述子 鲁棒性 匹配 

分 类 号：TP319.41[自动化与计算机技术—计算机软件与理论]