一种改进的Otsu红外林火图像提取方法研究  

An improved Otsu infrared fire image extraction method research

作　　者：王淑青[1] 朱道利 潘健[1] 李叶伟[1] 刘天俊 李维[1] 要若天 

机构地区：[1]湖北工业大学太阳能高效利用湖北省协同创新中心,武汉430068

出　　处：《激光杂志》2016年第10期99-101,共3页Laser Journal

基　　金：2015年国家级大学生创新创业训练计划项目(201510500008)

摘　　要：由于采集到的红外林火图像信噪比较低,现有的算法很难在提高信噪比的同时提取出红外目标图像。鉴于此,本文通过改进最大类间方差,构造合适的罚函数,将Otsu问题转化为求带约束条件的函数最优解问题,再通过迭代运算实现对林火目标的提取。实验结果表明,图像的噪声得到了有效地去除,同时较准确地提取出了林火目标,是一种比较稳健的目标提取算法。Due to the low forest fire infrared collected SNR, infrared image is extracted the target at the same time. In view penalty function, let the problem of Otsu transformed into existing algorithms is difficult to improve the SNR of of this, the paper uses the improved Otsu, tectonic seeking a function＇s optimal solution with the constraints, and then realize the extraction of forest fire goals by iteration. Experimental results show that the image noise has been effectively removed, while the target of forest fire is accurately extract, the algorithm is a more robust object extraction algorithm.

关 键 词：OTSU 红外林火 目标提取 罚函数 

分 类 号：TN209[电子电信—物理电子学]