基于OTSU分割的云层背景下弱目标检测算法研究  

The weak target detection algorithm under cloud background based on OTSU segmentation

作　　者：孙红光[1,2] 卜倩[1] 李欢利[3] 张瑾[1] 张慧杰[1] 

机构地区：[1]东北师范大学计算机学院,吉林长春130024 [2]长春理工大学机电工程学院,吉林长春130022 [3]中国科学院长春光学机械与物理研究所,吉林长春130033

出　　处：《东北师大学报（自然科学版）》2009年第2期79-83,共5页Journal of Northeast Normal University(Natural Science Edition)

基　　金：吉林省科技发展计划项目(20070322);东北师范大学自然科学青年基金资助项目(20081003)

摘　　要：在各种背景的弱目标检测算法研究中,采用了最大类间方差(OTSU)分割的检测算法,对于不同大小的目标提出了两种不同的处理方法.在背景简单的小目标的预处理中利用中值滤波和OTSU相结合的方法;对于背景相对复杂的大目标的检测采用自适应门限、拉普拉斯(Log)滤波和OTSU分割的检测算法,把目标提取出来.自适应门限用于增强图像,使图像的背景灰度变得均匀;Log高通滤波器可以有效地去除背景;OTSU是经典的非参数,无监督自适应阈值选取方法,对图像经过阈值分割后,图像将变成包含少量可能目标点的二值图像.仿真实验表明:该算法能够有效地去除背景天空的强浮云,具有计算量小等优点,能够很好地检测出目标.The detection of weak targets in the all kinds of background, in this paper, adopting OTSU segmentation method,and two different methods are put forward. The small targets pretreatment use of the median filtering and OTSU segmentation in simple background;the large target pretreatment use of the adaptive threshold, Log filtering and OTSU segmentation, the target can be extracted. Adaptive threshold used to enhance the image so that the gray background of the image becomes uniformly; the Log filter for the high-pass can remove the background effectively;the variance between the largest category is a non-classical parameters, unsupervised adaptive threshold selection method,after threshold segmentation of the image, the image will become a binary image which may contain the target. The simulation experiments show that the algorithm can effectively remove the background and has the small amount of computing advantages, which can inspect targets very well.

关 键 词：自适应门限 Log滤波 最大类间方差分割 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]