复杂场景下的自动红外舰船目标检测  

Automatic Infrared Warship Target Detection under Complicated Background

作　　者：邢莎 吉林[1] 雍杨[1] 胡俊杰[1] 邢根祥 

机构地区：[1]西南技术物理研究所,四川成都610041 [2]延川调频转播台,陕西延安716000

出　　处：《红外技术》2014年第4期320-325,共6页Infrared Technology

摘　　要：为实现多种复杂海天背景下红外舰船目标的有效检测,提出了一种基于局部边缘梯度特征分析的舰船目标检测方法。利用红外舰船目标图像的特征,改进自适应Butterworth高通滤波对图像进行预处理。通过计算局部边缘梯度特征和聚类分析确定初始目标区域,比较梯度特征参数并进行多尺度搜索准确检测出舰船目标。对多种复杂海天场景红外图像进行仿真实验,结果表明,该算法能够有效准确地检测出多种复杂海天场景下的红外舰船目标,且算法简单,易于实现,适应面广。In order to fulfill the warship target extraction of multiple complex scenes, this paper presents an algorithm based on gradient characteristic of infrared complicated sea-sky image. The infrared image was preprocessed by improved Butterworth high-pass filter which using features of infrared sea-sky image. Then the initial target＇s regional can be obtained by clustering the peak value of edge gradient. The correct warship was extracted by comparing the edge gradient parameters and multi-scale search. Experimental results indicated that this algorithm can detect the warship under various complicated sea-sky background effectively, and have good adaption, also can be implemented easily.

关 键 词：舰船目标检测 红外图像 梯度累加和 边缘梯度 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]