可调直方图均衡化的正则解释及其改进

摘要

将直方图均衡化看作典型的不适定问题,提出了可调直方图均衡化的正则化解释,以及为了获得较好的图像增强效果并给出其改进措施.首先,对可调直方图均衡化方法所对应的目标函数进行简化,得到了基于信息能量作为正则项的可调直方图均衡化崭新解释;其次,考虑到信息熵与信息能量之间呈对偶关系,于是获得了一种基于信息熵作为正则项的可调直方图均衡化新模型;再次,利用图像灰度级概率分布信息提出了可调直方图均衡化中正则项系数选取的方法;最后,为了改善可调直方图均衡化方法实现图像增强的效果,给出了基于灰度级累计概率修改的可调直方图均衡化改进方法.实验结果表明,可调直方图均衡化方法的修改是有效的.%In view of histogram equalization regarcled as typical ill-posed problem, the adjustable histogram equalization method is explaned by regularizaion theory, and its improvement is proposed in order to obtain the better effect of image enhancement. Firstly, the objective function of adjustable histogram equalization method is simplified and explaned by means of information energy as a regular item. Secondly,in consideration of the dual relationship between information entropy and information energy, the new adjustable histogram! equalization model based on information entropy as regularization is obtained. Thirdly, the regular item coefficient choosing method of adjustable histogram equalization is presented by rmeans of information entropy or energy of gray levels probablity distribution. Finally,the adjustable histogram equalization improvement method based on modifying cumulative probability of gray levels is proposed in order to improve the image enhancement effect of traditional histogram equalization method. Experimental results show that the improvement method of adjustable histogram equalization is effective.