摘要
针对传统模板匹配中矩形模板计算量大、容易造成误匹配的问题,对矩形模板进行改进,提出了十字模板匹配算法;对待匹配的两幅图像进行边缘提取,结合两幅图像的灰度图像获得灰度边缘图像,并利用十字模板匹配方法匹配选定的模板区域;最终在连续性约束的条件下,利用欧氏距离完成特征点匹配。对比实验结果表明十字模板匹配速度快,准确度高;基于十字模板的特征点匹配方法简单可靠,具有一定的实用性。
According to the computation complexity and liability to make the matching problem of the rectangular template in the traditional template matching,the improvement is made to the rectangular template and a cross template matching algorithm is presented; edges are detected on the two images to be matched, and the two gray images are combined.Gray edge images are obtained.Then the selected template area is matched by the cross template matching method;finally,feature points are matched by the Euclidean distance under the continuity constraint.The results of comparative tests show that the cross template is fast and has high accuracy; feature points matching method based on the cross template is simple and reliable, and has certain practicability.
出处
《计算机工程与应用》
CSCD
北大核心
2010年第29期167-169,共3页
Computer Engineering and Applications
基金
国家林业局"948"项目(No.2010-4-05)
黑龙江省科技攻关项目(No.GB08A502)
东北林业大学研究生科技创新项目~~
关键词
双目立体视觉
特征点匹配
测距
binocular stereo vision
feature points matching
ranging
