基于多特征整合的双色红外小扩展目标精确分割算法  

The Algorithm of Two Color IR Small Extended Target Precision Segmentation Based On Multiple Features Integration

作　　者：李秋华[1] 杜鹢[2] 

机构地区：[1]解放军重庆通信学院军事信息工程系,重庆400035 [2]通信网络技术管理中心,北京100840

出　　处：《红外技术》2009年第2期112-118,共7页Infrared Technology

基　　金：国家高技术研究与发展计划(973计划)资助项目(No.2007CB311003)

摘　　要：针对复杂背景下红外图像小扩展目标分割的问题,提出了一种基于多特征整合的双色红外小扩展目标精确分割算法。该算法在提取目标在红外双波段图像中的多种图像特征的基础上,采用D-S证据理论中的正交和复合规则对来自两个不同红外波段的目标检测证据进行综合处理,充分利用了目标在中波和长波红外图像的信息互补性和冗余性,在较大程度上提高了杂波背景下红外图像小扩展目标定位与分割的稳健性和精确性。实验结果显示了该算法的有效性。Aim at the problem of IR image small extended target segmentation under complex background, this paper presented an algorithm for two color IR small extended target precision segmentation based on multiple features integration. The algorithm integrated these target detection evidences from the two different IR bands using the orthogonal sum rule in the D-S evidential theory after extracting the multiple image features of these potential targets in the IR dual band images, take full advantage of the complementary and redundancy of target＇s information in the middle wave and long wave IR images; and improved the robustness and precision of IR image small extended target location and segmentation under highly cluttered background. The experimental results show the effectiveness of this algorithm.

关 键 词：双色红外 信息融合 证据理论 多特征整合 目标分割 

分 类 号：TN957.51[电子电信—信号与信息处理]