摘要
针对图像增强过程中,分数阶微分的阶数往往由经验或大量的实验来选择较优的值,不能实现自适应性,没有充分发挥分数阶微分的优良特性的问题,提出了一种基于图像局部梯度、信息熵和方差三个与图像纹理相关的参数的自适应分数阶微分图像增强算法,并应用于一些相关的医疗图像中。依据信息熵和平均梯度等纹理分析的定量评定标准,对增强后的图像进行实验比较分析。实验结果表明,相对于所比较的算法,自适应分数阶微分算法能够在增强图像的边界和纹理部分的同时,保留平滑区域的信息细节,同时获得较好的视觉效果。
In the process of image enhancement,the optimal value of the fractional order was often chosen by the experience or a large number of experiments,it couldn't realize the adaptation,and it did not give full play to the excellent characteristics of the fractional differential. This paper proposed an adaptive fractional order differential image enhancement algorithm based on 3 parameters related to image texture,which were the local gradient of the image,the information entropy and the variance. Moreover,this proposed algorithm was applied to some related medical images. Based on the quantitative assessment criteria of texture analysis,such as information entropy and average gradient,this paper analyzed and compared the enhanced image. Experimental results show that compared with other algorithms,adaptive fractional order differential algorithm can not only enhance the image edge and texture,but also keep smooth regions of details and gain better visual effect.
出处
《计算机应用研究》
CSCD
北大核心
2017年第12期3895-3898,3903,共5页
Application Research of Computers
基金
重庆市基础科学与前沿技术研究资助项目(cstc2015jcyj BX0124)
关键词
图像增强
分数阶微分
梯度
信息熵
方差
自适应
医学图像
image enhancement
fractional order differential
gradient
information entropy
variance
adaptive
medical image
作者简介
陈向阳（1965-）,女,讲师,主要研究方向为微分方程及其应用;;谭礼健（1990-）,男（通信作者）,硕士研究生,主要研究方向为数字图像处理（857116721@qq.com）.
