图像点集的不变性匹配方法

THE METHOD OF MATCHING WITH INVARIABLE PORPERTY FOR SETS OF IMAGE POINTS

许多图像目标都可以用其上有代表性的一些点(如角点或边缘上的点)及其相互关系作为特征进行描述,通过模板点集和目标点集之间的匹配可以达到识别的目的。本文提出了一种新的平面点集间的与各自的平移、旋转、尺度变化和排列次序无关的距离描述,这种描述在一定范围内具有唯一性,从而为以点集特征匹配为基础的识别提供了一种点集之间差别有效的衡量手段。

Many objects in images can be present by some important points on them (e.g. corner points or edge points). We could complete recognition task through matching these points. In this paper, we proposed a new definition of the distance between two sets of image points, this definition have invariable property for ratatin, translation, variance of scale and the order of alignment of this points, and the distance is exclusive in some ranges. So a new effective means is given for the recognition that based on the set of image points.