基于分形的人造目标与自然物体区别  

Recognition of Manmade Target and Natural Object Through Fractal

作　　者：曾鹏鑫[1] 么健石[2] 朱琳琳[1] 徐心和[1] 

机构地区：[1]东北大学教育部暨辽宁省流程工业综合自动化重点实验室,辽宁沈阳110004 [2]沈阳炮兵学院自行火炮系,辽宁沈阳110162

出　　处：《东北大学学报（自然科学版）》2006年第3期260-263,共4页Journal of Northeastern University(Natural Science)

基　　金：国家自然科学基金资助项目(60475036);国防预研项目

摘　　要：提出一种基于物体分形特征的人造目标与自然物体的区分方法.该方法根据人造目标和自然物体的固有差异,将分形维数作为估计表面粗糙度的一个重要参数,结合“毯子”维算法来计算物体表面粗糙度.该算法快速、简单、有效.实验表明,人造目标的分形维数较低,自然目标的分形维数较高,用此算法计算物体分形维数区别人造目标和自然物体是可行的.Based on object＇s fractal dimension, an approach was proposed to differ manmade target from natural objects. According to the innate difference between manmade targets and natural objects, fractal dimension was taken as a very important parameter to estimate the surface roughness, and the surface roughness of the target was calculated by the algorithm proposed by Peleg et al. The method is fast, simple and effective. Experimental result showed that the number of fractal dimensions of manmade targets is lower than that of natural objects. It is feasible to differ manmade targets from natural objects by the fractal dimension calculation.

关 键 词：分形 分形维数 模式识别 图像处理 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]