运用隔帧模板匹配算法检测复杂背景下运动目标  

Using Discontinuous Frame Template Matching Algorithm to Detect Moving Target Under Complex Background

作　　者：强世锦[1] 申金娥[2] 荣健[2] 

机构地区：[1]武汉职业技术学院通信工程系,湖北武汉430074 [2]电子科技大学物理电子学院,四川成都610054

出　　处：《红外技术》2008年第1期27-30,共4页Infrared Technology

基　　金：国家自然科学基金项目(编号：60572079)

摘　　要：提出一种用于复杂背景下实时识别运动目标的新方法,该方法首先利用预处理滤除大部分灰度较低的背景和杂波,然后再采用不同间隔的序列模板分别进行二次误差匹配并构成并行结构来识别目标。该方法成功地克服了传统的模板匹配法计算量大,耗时多的不足,并使整个系统能够探测到静止甚至速度较大的运动目标,且在部分被遮挡甚至偶尔目标失锁的情况下依然能较好地跟踪。仿真结果也表明了该方法的有效性。A new detection method used in real-time recognizing moving target under complex background is presented. Firstly, pretreatment is used to restrain most of the background clutter. Secondly, different internals templates are employed in twice error matching process and form a parallel structure in targets recognizing. The method successfully overcomes complex calculation, time-consuming and inadequate in traditional template matching, it makes entire system detect stationary or moving targets, it can have a good tracking when the target was partially obscured and even blocked occasionally,. The simulation results indicate that the method is feasible.

关 键 词：对比度 隔帧模板匹配 红外图像 二次误差匹配 

分 类 号：TP75[自动化与计算机技术—检测技术与自动化装置]