摘要
为解决图像的精确配准问题,提出了结合LoG算法的特征点的提取方法,并将尺度不变特征算法(SIFT)应用到图像的特征描述中.首先利用LoG算法计算边缘点,对边缘点的梯度值进行排序,选择梯度较大的点作为特征点;然后采用SIFT计算特征点的特征向量,利用最小距离算法找到两幅图像的匹配点对;最后利用最相关点和次相关点比例的方法排除错误的点对.实验结果证明,算法对具有光照、角度不同的两组图像能够实现精确的配准,准确率超过90%.
The feature point extraction method is proposed combining with LoG algorithm,and Scale Invariant Feature Transform algorithm is applied to image feature extraction.First of all,LoG algorithm is used to compute the edge points,the edge points are sorted by the gradient values,and the points of lager gradient are selected as feature points;Then compute the feature vector of feature points with SIFT algorithm,and the minimum distance algorithm is used to find the matching points of two images;Finally,the wrong matching points are excluded by the ratio algorithm of the most relevant point and the second relevant point.The experimental results prove that the method proposed in this paper can achieve accurate matching for the two images with different illumination and angle,and the accuracy was over 90%.
出处
《小型微型计算机系统》
CSCD
北大核心
2012年第11期2535-2537,共3页
Journal of Chinese Computer Systems
基金
国家自然科学基金项目(60871093)资助
浙江省教育厅科研项目(Y200907801)资助
关键词
图像匹配
边缘检测
点特征
SIFT
image matching
edge detector
point feature
scale invariant feature transform
