一个高性能的自动目标识别算法  

A high performance automatic target recognition algorithm

作　　者：陈东[1] 刘文华[2] 李飚[1] 沈振康[1] 

机构地区：[1]国防科技大学ATR国家重点实验室,长沙410073 [2]中央电视台,北京100859

出　　处：《计算机工程与设计》2001年第4期86-88,共3页Computer Engineering and Design

基　　金：中国博士后科学基金; 国防科技重点实验室基金(347A1)资助

摘　　要：提出了一个高性能的自动目标识别算法，该算法基于遗传算法，在算法中需要提取目标的位移、旋转和比例不变特征。算法具有较强的容错性和抗干扰能力，不仅能够对各类形状有一定差异的目标进行分类，而且对部分缺损或有噪声污染的目标也能够正确进行分类。实验结果证明了算法的有效性。This paper describes an automatic target recognition algorithm based on genetic algorithm. The targets are rotated, translated, and scaled in this algorithm. This algorithm has highly robust performance and error-tolerated ability. It has not only classified to those targets which are different in shape to a certain extent, but also classified to those targets which are partially distorted and polluted by noise. The experiment results show the algorithm is available.

关 键 词：遗传算法 自动目标识别 不变性特征 图像理解 计算机视觉 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]