天空区域图像的增强算法的改进  

Improvement of enhancement algorithm for aerial image

作　　者：邓玥[1,2] 王延杰[1] 李静宇[1,2] 陈晓曦[1,2] 张叶[1] 

机构地区：[1]中国科学院长春光学精密机械与物理研究所,吉林长春130033 [2]中国科学院研究生院,北京100039

出　　处：《激光与红外》2012年第9期1080-1085,共6页Laser & Infrared

基　　金：国家自然科学基金青年基金(No.60902067)资助

摘　　要：传统的子块部分重叠直方图均衡化(POSHE)算法在增强图像对比度的同时会在灰度变化缓慢的区域造成过增强效果,产生块状效应和噪声,严重影响图像质量,本文针对这一缺点对该算法进行改进,将图像中灰度变化缓慢的天空区域从图像中分离出去,对分离出的天空区域进行全局直方图均衡化处理,对分离后的图像进行局部直方图均衡化处理。通过Matlab仿真实验可以证明本文算法不仅可以增强图像的对比度,还可以加强图像的整体和细节信息,有效防止过增强并抑制噪声,使图像的质量和视觉效果都得到提高。The traditional Partially Overlapped Sub-block Histogram Equalization may cause over enhancement and block effect to the sky region image with slow gray gradient, which degrades the quality of the image seriously. To solve this problem, this article will improve the algorithm by separating the sky region with slow gray gradient from the im- age, and then deal the separated aerial aera with the Overall Histogram Equalization while disposing the remainder aera with Partially Overlapped Sub-block Histogram Equalization. Experiment results with Matlab reveal that the algorithm proposed in the article can not only enhance the contrast of the image but also improve the information in both overall image and partial image. It can also avert over-enhancement and repress noise simultaneously, which highly improves the quality and visual effect of the image.

关 键 词：图像增强 直方图 天空区域 阈值分割 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]