基于特征匹配的目标实时识别  

A REAL TIME OBJECT RECOGNITION SYSTEM BASED ON FEATURE MATCHING

作　　者：徐正伟[1] 郑兰芳[1] 吴成柯[1] 陆心如[1] 

机构地区：[1]西安电子科技大学信息工程系,710071

出　　处：《自动化学报》1993年第4期428-432,共5页Acta Automatica Sinica

基　　金：电子预研基金

摘　　要：本文介绍了一个基于特征匹配的运动目标实时识别系统。该系统能对二十九种以上不同形状的目标或十一种形状相似的飞机模型实现正确识别,而且其识别结果与目标的平移、比例缩放及旋转无关,正确识别率高于98％,识别时间小于34ms,满足实时要求。A real time object recognition system based on feature maaching is described in this paper. The system can recognize more than 29 object models or more than 11 plane medels with similar shapes. The results of recognition are independent of translation, rotation and scale. The experiments show that the corrvect recognition rate is greater than 98％, and the recognition time is less than 34 ms which meet the need for recognizing the moving objects in real time.

关 键 词：目标识别 图像处理 匹配 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]