基于模糊积分的可信度融合目标检测  

Confidence fusion targets detection based on fuzzy integral

作　　者：熊大容[1] 杨烜[2] 

机构地区：[1]深圳大学计算机与软件学院,广东深圳518060 [2]深圳大学信息工程学院,广东深圳518060

出　　处：《激光与红外》2009年第6期655-659,共5页Laser & Infrared

基　　金：深圳市科技计划项目资助

摘　　要：针对可见光和红外热图像序列中的远距离目标检测问题,提出了一种基于模糊积分的可信度融合的目标检测方法。该方法通过帧间差累积,在两种传感器的图像中确定了运动目标区域,根据目标区域的强度分析,定义了目标可信度度量,进而利用模糊积分融合函数进行可信度融合以实现目标检测。试验结果表明模糊积分分类效果较好,从而证明了该方法在目标识别中的可靠性和可信度,具有一定的实用性。Aimming at long distance targets detection problem in visual and thermal infrared image sequences, a confidence fusion detection method based on fuzzy integral. The method acquires the moving targets area by frame difference accumulation in two sensor image sequences, and we defined the target reliable measurement by the intensity analysis of the moving target area. The detection results are acquired by confidence fusion using the fuzzy integral fusion function. The experimental results show that fuzzy classification effectiveness better in order to illustrate that this approach to detect the target is reliable and robust.

关 键 词：可见光图像序列 红外图像序列 模糊积分 可信度融合 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]