复杂背景下目标的快速识别方法  

The fast identification method of target under complex bockground

作　　者：张会森 

机构地区：[1]长庆油田公司培训中心,西安710127

出　　处：《科技与企业》2014年第7期324-325,共2页Science-Technology Enterprise

基　　金：国家自然科学基金重点项目(No.60736008)

摘　　要：复杂背景下目标的识别是自动跟踪测量系统中的关键技术,识别算法的优劣直接影响了目标识别的精度与速度。针对空中目标成像,由于大气湍流等因素的影响,CCD图像中的光斑通常呈不规则形状,而传统的模板匹配法与圆拟合法在识别不规则光斑时存在不足。本文利用光斑在像平面上的空域分布特性,提出一种基于空域相关程度排序的识别算法,配合经过改进的高效图像处理方法,实现了多个不规则光斑的快速识别。实验结果表明该方法能够在混有多个假目标的图像中有效识别出真实目标并进行跟踪测量。The fast identification method of target under complex bockground is the key technology of automatic tracking measurement system.The quality of recognition algorithm directly affected the target recognition speed and accuracy.Aimed at the aerial target imaging, since atmospheric turbulence factors,the spots of CCD image usually appears irregular shape, the traditional template mat ching method and the light fitting method are not very effect when these spots exist.In this paper,we Proposed an recognition algorithm based on sorting space-domain relevance. Cooperate with efficient image processing method,this algorithm achieved quick identification of multiple irregular spot. Experimental results show that the proposed method can effectively recognize the true target ,moreover tracking and measure this target in the image mixed with several false target.

关 键 词：测量技术 目标识别 光斑质心 迭代搜索 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]