一种邻近性的计算方法

摘要

邻近律是Gestalt定律的基本定律之一，然而迄今为止，如何在图像处理中计算元素之间的邻近性程度还没有具体方法提出．以点元素作为抽象载体提出了一种邻近性的计算方法，该方法借鉴基于最小支撑树（MST）的聚类思想，引入三角剖分和Prim算法计算元素之间的MST邻近性测度和加权MST邻近性测度用于度量元素之间的邻近性．实验结果表明．这一计算方法与视觉感知结果一致，而且可以给出元素之间邻近性程度更为精细的划分．%The proximity law is one of the basic Gestalt law. However, the computation problem of measuring the proximity extent of image elements still remain unsolved. Inspired by the cluster algorithm called MST （Minimal Spanning Tree）, this paper designs a computation method for proximity. Taking the image elements as points or lines, the method introduces the delaunay triangularization and Prim algorithm to get the MST proximity measure and the weighted MST proximity measure to indicate the proximity extent. Experimental results show that the computation results are consistent with human perception and can give a more fine distinction to different proximity extent.