复杂背景下的最大熵阈值红外目标分割算法  

Infrared target segmentation algorithm based on maximum entropy threshold in complex background

作　　者：温凯峰[1] 叶仕通[2] 万智萍[3] 

机构地区：[1]嘉应学院,广东梅州514015 [2]广东工业大学华立学院,广东增城511325 [3]中山大学新华学院,广东广州510520

出　　处：《激光与红外》2016年第1期103-108,共6页Laser & Infrared

基　　金：广东省自然科学基金项目(No.S2013010013307);广东省青年创新人才自然科学类基金项目(No.2014KQNCX251);广东省青年创新人才自然科学类基金项目(No.2014KQNCX253)资助

摘　　要：为了有效地分割出实际拍摄中的红外目标信号,提出了一种适用于复杂背景下的最大熵阈值红外目标分割算法;该算法根据图像中信号的相关性,对图像的干扰信号的预处理,随后对所得的图像信号进行熵值的求取;接着根据目标与背景信号的分布特性,对所获得的目标信号与背景信号进行深入提纯处理,最终获得更为理想的红外目标信号。实验结果表明,该算法能够在不同背景下始终保持较好的红外目标分割效果,且具有耗时低、清晰度高的分割优势。In order to effectively segment the infrared target signal of actual shooting, an infrared object segmentation algorithm based on maximum entropy threshold is proposed in a complex background. According to the correlation of image signals,interference signals of the images are preprocessed, and then the entropy in those image signals is calculated. After that, in order to obtain more desirable infrared target signal, the target signal and the background signal are further processed according to the distribution eharacteristies of the target and the background signal. The experimental results show that this algorithm can always maintain good infrared target segmentation in different backgrounds. And it has the advantages of low consumption and high definition.

关 键 词：红外图像 分割算法 阈值 相关性 目标提纯 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]