种结合小波变换的SIFT特征图像匹配算法  

Image Matching Algorithm by Wavelet Transform Combined with SIFT Characteristics

作　　者：刘佳嘉[1] 何小海[1] 陈为龙[1] 

机构地区：[1]四川大学电子信息学院,四川成都610064

出　　处：《计算机仿真》2011年第1期257-260,335,共5页Computer Simulation

摘　　要：研究目标识别和图像匹配优化问题,针对小波变换能压缩图像信息及抑制高频噪声的优点,采用小波变换与尺度不变特征变换相结合的图像匹配方法,克服了传统SIFT算法计算复杂实时性不强的缺点。算法首先对基准及后续图像分别做小波分解得到包含大量信息的低频图像,利用SIFT提取特征点,通过优化的最优节点优先(BBF)算法及随机抽样一致性(RANSAC)算法完成两图特征点对的搜索及检验。实验表明,算法不仅减少了匹配计算量,大大提高匹配效率,同时对图像缩放,旋转,遮挡,亮度变化也有较强的适应能力。In order to overcome the available disadvantage of the complicated computation and bad real time fea- ture of SIFT, a method of image matching is proposed by combining wavelet transform with Scale Invariant Feature Transform, with the advantage of the capability of data compression and noise reduction of wavelet transform. The low -frequency image contained a large amount of information can be obtained through wavelet decomposition from both the original image and the model image. And then features are detected using SIFT algorithm. The matches are con- structed by the optimized implementation of BBF algorithm, and verified by RANSAC. The algorithm can reduce computation cost of image matching and improve the speed, besides, it is invariant to image scale, rotation, cover and il- lumination change.

关 键 词：多小波变换 图像匹配 尺度不变特征变换 特征点提取 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]