基于Otsu准则的红外图像快速分割算法  

Fast infrared image segmentation algorithm based on Otsu criterion

作　　者：王宁 周铭[1] 杜庆磊[1] 王冰 WANG Ning;ZHOU Ming;DU Qinglei;WANG Bing(Air Force Early Warning Academy,Wuhan 430019, China)

机构地区：[1]空军预警学院,武汉430019

出　　处：《空军预警学院学报》2019年第2期88-92,共5页Journal of Air Force Early Warning Academy

摘　　要：为实现红外图像快速目标检测,提出了一种基于Otsu准则的红外图像快速分割算法.首先运用模糊算法调整全局滤波的权值进行预处理,利用像素点的二维信息直接建立阈值直线的截距直方图;然后根据Otsu准则求解最佳截距阈值,使用该阈值直接对含二维信息的图像进行分割;最后通过自适应参数的形态学操作提升图像质量.实验结果表明,与二维直分法和Otsu直线阈值法相比,本文算法分割精度较好,具有噪声抑制能力,占用较少内存空间,对多目标图像具有较好的分割效果.In order to realize fast target detection in infrared image, this paper proposes a fast infrared image segmentation algorithm based on Otsu criterion. Firstly, the paper uses fuzzy algorithm to adjust the weight of global filtering for preprocessing, and directly establishes the histogram of the intercept of a threshold line by using the two-dimensional information of the pixel points. Then the paper solves the optimal intercept threshold according to the Otsu criterion, and also directly uses the threshold to segment the image with two-dimensional information. Finally, the image quality is improved by adaptive parameters morphological operations. Experimental results show that compared with the two-dimensional direct segmentation method and Otsu linear threshold method, the proposed algorithm has better segmentation accuracy, with noise suppression ability, less memory space taken up and better segmentation effect for multi-target images.

关 键 词：红外图像 图像分割 OTSU准则 阈值选取 模糊算法 

分 类 号：TN911.73[电子电信—通信与信息系统]