基于边缘特征的水下目标分类识别算法  

Classification and Recognition of Underwater Target based on Edge Feature

作　　者：陈立强[1] 刘忠[2] 田晓东[2] 

机构地区：[1]海军装备研究院系统所,北京100011 [2]海军工程大学电子工程学院,湖北武汉430033

出　　处：《自动化技术与应用》2007年第8期77-79,共3页Techniques of Automation and Applications

摘　　要：在水下声纳图像目标检测与识别技术中,水雷等目标的识别问题是关键技术之一。基于形状相似度的概念,利用目标边缘轮廓特征,给出一种基于距离多集方法的水雷目标形状分类算法。首先对检测到的水雷目标阴影区域进行规格化处理,采用距离多集方法对常见水雷形状进行分类识别,提出了一种模板更新策略。通过对球形、圆柱形和圆台形三类目标的阴影区域进行仿真分类计算可知,该方法具有仿射不变性,分类准确率较高。This paper presents a novel algorithm for mine shape classification by using edge feature of target based on the concept of shape similarity.The shadow region of mine is first extracted and normalized,then the distance multi-set method is used to classify the contour of mine targets.A Template renewing method is presented. Simulations with spheric,column and taper targets show that the correct classification ratio is 92.7%.This method is affine-transform invariant and high orecision.

关 键 词：边缘 距离多集 形状相似度 目标识别 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]