基于Trous算法的红外序列图像中运动目标的检测与跟踪  

Moving Target Detection and Tracking in IR Image Sequences Based on Trous Algorithm

作　　者：徐永兵[1] 

机构地区：[1]华中科技大学计算机学院,湖北武汉430074

出　　处：《红外》2015年第1期35-39,共5页Infrared

摘　　要：主要研究了红外图像中运动点目标的检测及跟踪问题。在复杂背景下,根据小目标在红外影像中容易丢失的特性,利用trous小波变换对图像非抽取(变换后图像数据未损失)的优点对红外图像进行了背景抑制与增强。通过聚类分割算法分割出了可疑目标,并采用八邻域分析算法准确跟踪并提取出了运动小目标。试验结果表明,本文算法在对红外序列图像中的运动目标进行检测与跟踪时具有较强的实用价值。The detection and tracking of moving point targets in infrared images are studied mainly. Background suppression and enhancement are carried out for the infrared images with complex back- ground. These processing are implemented by utilizing the advantages of Atrous wavelet transform according to the feature that small targets are easy to be lose in infrared images. The possible targets are segmented by using a Fisher algorithm. Then, the small moving targets are tracked and extracted accurately by using an Eight Neighboring Region Analysis Algorithm （ENRAA）. The experimental result shows that the proposed algorithm is of important practical value to the detection and tracking of moving targets in infrared image sequences.

关 键 词：ATrous小波变换 图像分割 红外运动目标 八邻域分析 目标检测和跟踪 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]