摘要
在分析了经典SIFT算法的基础上,提出了一种基于Canny算子和K-L变换的改进SIFT匹配算法。该方法首先利用Canny边缘检测算法获得图像的边缘点坐标,与SIFT算法检测出图像关键点的坐标进行对比以去除不稳定的边缘点;其次通过K-L变换,将特征描述符进行降维处理,降低算法复杂度;最后使用RANSAC算法剔除误配点。通过实验表明,该算法能有效去除不稳定的边缘响应特征点,减少图像匹配时间,提高图像匹配的准确性和鲁棒性。
Based on the analysis of traditional sIFr algorithm, an improved SIlT algorithm use canny operator and K-L transformation is proposed. Firstly, the canny edge detection is used to require the edge coordinates of image, this points are contrasted with the coordinates of the key points detect by traditional SIlT. Secondly, K-L transformation is used to reduce the feature descriptor dimension and complexity of the algorithm. Finally, the RANSAC algorithm is applied to remove mismatching points. The experimental outcomes display that the improved algorithm can remove unstable edge points ,decrease the matching time, increase the accuracy and robust of image matching.
出处
《电视技术》
北大核心
2014年第15期61-64,75,共5页
Video Engineering
基金
国家自然科学基金项目(61162020)