基于两层策略的噪声图像对的点匹配算法  

Point Matching Algorithm for Image Pairs with Noise Based on Two-Level Strategy

作　　者：栾广宇[1] 叶东[1] 车仁生[1] 

机构地区：[1]哈尔滨工业大学自动化测试与控制系,黑龙江哈尔滨150001

出　　处：《计算机仿真》2010年第6期254-257,共4页Computer Simulation

基　　金：国家自然科学基金(50275040)

摘　　要：针对机器人视觉研究中噪声的存在,很多的区域匹配算法都受其影响。为了解决去噪问题,提出了一种两层策略的噪声图像对的点匹配算法。先粗糙匹配采用较大的模板窗利用差值平方和比较边缘相似性,从而得到匹配点的大致范围。然后精细匹配,在粗略匹配所得的小范围内采用灰度相似性确定匹配点位置。结合数学组合的定义和噪声的对称概率密度函数,改进次序统计滤波器并且把它用于估计灰度值。以添加高斯噪声的真实图像为测试对象。实验结果显示了此匹配在不同的信噪比下受噪声影响很小和具有很高的匹配率。Many area -based methods are affected by noise. To solve this problem, a new point matching algorithm based on the two -level strategy is presented. Firstly, sum of squared difference （SSD） is used to measure the edge similarity within a large template window to obtain a coarse level matching of points based on the contours of the images. Then gray similarities of the points are measured within a small region in the fine - level matching based on the result of the coarse - level matching and the correspondences of the points are localized accurately. Combining the combination definition and symmetrical probability density function of noise, the order statistic filter （OSF） is improved and the intensity is estimated by the improved OSF. The algorithm is applied to real images corrupted with Gaussian noise. It is shown that the matching is little affected by image noise under different signal to noise ratio levels and correct rate of its matching is high.

关 键 词：噪声 图像匹配 改进的滤波器 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]