改进SURF算法的特征提取与匹配方法研究

Research on Feature Extraction and Matching Method of Improved SURF Algorithm

针对门把手图像特征点提取与匹配对快速性和准确性的要求,提出一种改进SURF算法。该算法主要对图像较平滑区域难以提取出大量信息点的问题进行改进。增加了边缘检测算法,得到图像边缘信息后进行形态学处理,并通过膨胀运算和开运算后获得门把手图像边缘区域信息并提取出关键点,进而获得较多特征明显的信息。改进SURF算法和原始SURF算法相比,在平滑区域能够较好提取出特征点,匹配准确率也有明显的提升,并且增强了算法的实时性。

To meet the requirement of fastness and accuracy of doorknob image feature extraction and matching,an improved SURF algorithm is proposed,which mainly targets the problem that it is hard to extract a large number of information points at the smooth area of the image.In addition,the edge detection algorithm is added.With morphological processing,dilation operation and open operation,the image edge information of doorknob was obtained and the key point was extracted,hence some more featured information were collected.The results show that the improved SURF algorithm can extract feature points in the smooth area in real time with obvious improvement in the accuracy rate of matching.