基于迭代阈值的子块部分重叠双直方图均衡算法  

Bi-histogram Equalization Based on Partially Overlapped Sub-block with Iterative Threshold

作　　者：孙水发[1] 李乐鹏[1] 董方敏[1] 邹耀斌[1] 陈鹏[1] 

机构地区：[1]三峡大学智能视觉与图像信息研究所,湖北宜昌443002

出　　处：《广西师范大学学报（自然科学版）》2013年第3期119-126,共8页Journal of Guangxi Normal University:Natural Science Edition

基　　金：国家自然科学基金资助项目(61102155;61272237;61272236);湖北省高等学校优秀中青年科技创新团队计划项目(T201002)

摘　　要：双直方图均衡通过选取适当的阈值将图像分为两个子图再均衡可以有效避免直方图均衡所产生的灰度级简并和过增强等现象,但阈值的选取是关键。基于迭代阈值的双直方图均衡算法(IBBHE)通过迭代方法选取双直方图的分割阈值,不仅保持了图像的亮度信息,还避免了图像过度增强现象的出现。本文在此基础上,将IBBHE的迭代阈值选取思想引入部分重叠的局部直方图均衡算法(POSHE),提出一种改进的基于迭代阈值的子块部分重叠双直方图均衡算法(IPOSBHE)。针对偏暗和偏亮的图像处理实验结果表明:该算法既能有效地增强图像的细节信息,又能保持输入图像的亮度,使均衡后的图像具有更加自然的视觉效果。Histogram equalization （HE） makes the average of equalized image close to gray median re-gardless the original image. The gray levels of HE are degenerated and the detail of the image is lost,and the over enhancement and unnatural effect is occurred easily. For bi-histogram equalization,original im-age is decomposed into two sub-images based on an appropriate threshold and the sub-images are equal-ized independently. However, the threshold is the key issue. The iterative method for choosing the threshold of the bi-histogram is proposed previously,i, e. iterative threshold based brightness preserving bi-histogram equalization （IBBHE）. Experimental results on dim medical image show that this method not only keeps the image brightness ,but avoids the phenomenon of over enhancement. Based on partially overlapped sub-block histogram equalization algorithm （POSHE）,an iterative threshold based partially overlapped sub-blocks bi-histogram equalization algorithm is proposed （IPOSBHE） and the iterative method is adopted in the process of sub-block histogram equalization. Experimental results on both dim medical image and bright scene image show that the local image details are enhanced ,and the input image brightness is also preserved effectively. The blocking effect on the image is eliminated and the equalized image shows a more natural visual impression.

关 键 词：图像增强 直方图均衡 双直方图均衡 局部直方图均衡 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]