基于自适应直方图规定化函数引导的动态分层图像增强算法

Dynamic partition image enhancement algorithm induced by optimal adaptive histogram specification function

现有直方图规定化技术存在两方面不足:无法自适应构造一个能突出原图特征的直方图规定化函数,常需要人为指定,使增强效果难以控制;增强对象缺乏目的性,难以满足感兴趣灰度区域的特殊要求。针对该问题,本文提出基于自适应直方图规定化函数引导的动态分层图像增强算法。首先通过分析图像直方图灰度分布来分割灰度分布子图,并根据用户感兴趣灰度确定各子层灰度映射范围;然后利用各子层直方图自适应地构造能突出子层图像直方图特征的目的直方图规定化函数,并构建映射函数,将子层灰度映射到指定范围;最后综合各增强结果得到最后增强图像。实验结果表明,本算法能自适应构造突出原直方图特征的规定化目的直方图,又对用户感兴趣灰度有目的地进行控制增强,使所增强图像具有更加合理的视觉效果。

Existed histogram specification technology has two drawbacks although it makes the enhanced image have a high contrast. It could not adaptively construct an optimal histogram mapping function which can emphasize the characteristics of original histogram, usually need to define by user, so the effect of enhancement is difficult to control. In addition, the enhanced object lacks purpose, and it is difficult to meet the special requirements such as the region of interest. To solve this problems, dynamic partition image enhancement algorithm induced by optimal adaptive histogram specification function is proposed in this paper. First images are separated by analyzing the image histogram, according to the interested in gray range confirm each sub-level gray range of mapping. Then sub-level histogram is used to adaptively construct a desired histogram which can emphasize the characteristics of sub-level histogram, and histogram specification algorithm is adopted to construct gray transfer function, then sub-level gray is mapped to the appointed range. Finally the enhancement result of sub-level is synthesized to obtain the last enhancement image. The experimental results demonstrate that dynamic partition image enhancement algorithm induced by optimal adaptive histogram specification function not only can adaptively construct a optimal desired histogram which can emphasize the characteristics of original histogram, but also can control the region of interested. This proposed algorithm makes the enhancement image have a more reasonable visual effect.