自适应分数阶微分的图像增强及应用
Image enhancement and application based on adaptive fractional order differential
作者：张玉(四川大学电子信息学院)；王正勇(四川大学电子信息学院)；滕奇志(四川大学电子信息学院)；袁晓(四川大学电子信息学院)
Author：ZHANG Yu(College of Electronics and Information Engineering, Sichuan University)；WANG Zheng-Yong(College of Electronics and Information Engineering, Sichuan University)；TENG Qi-Zhi(College of Electronics and Information Engineering, Sichuan University)；YUAN Xiao(College of Electronics and Information Engineering, Sichuan University)
收稿日期：2014-03-19 年卷（期）页码：2015,52(1):93-100
期刊名称：四川大学学报: 自然科学版
Journal Name：Journal of Sichuan University (Natural Science Edition)
关键字：自适应函数; 和声搜索算法; 信息熵; 平均梯度; 岩屑识别
Key words：The adaptive function; Harmony Search algorithm; Information entropy; Average gradient; Debris recognition
基金项目：国家自然科学基金（60972130,6372174）
中文摘要
为了改善图像质量, 本文提出了自适应分数阶微分的图像增强算法.该算法首先在Grünwald Letnikov分数阶微分定义的基础上, 充分利用相邻像素点的信息, 设计了具有旋转不变性的分数阶微分掩模; 其次, 利用图像梯度, 局部的信息熵和对比度构造分数阶微分阶次的函数, 利用和声搜索算法选取最优参数, 建立了图像增强和分数阶微分阶次的非线性量化关系; 最后采用信息熵、平均梯度对增强后的图像定量分析, 为验证该算法的有效性, 把用该算法增强后的岩屑图像应用在岩屑识别中.实验结果表明, 能够实现图像的自适应纹理细节信息增强, 增强效果明显; 同时岩屑识别率有一定的提升作用.
英文摘要
In order to improve the image quality, an adaptive image enhancement algorithm of fractional order differential is proposed. Firstly, according to the definition of fractional order differential based on Grünwald Letnikov, and making full use of the information of adjacent pixels, rotation invariance of the fractional order differential mask is designed. Secondly,according to the image gradient, local information entropy and contrast, fractional order differential order function is constructed.The optimal parameters are selected by Harmony search algorithm and nonlinear quantitative relationship of image enhancement and the fractional order differential order is established.Finally,information entropy and average gradient are used to analyze the processed image quantitatively.To verify the effectiveness of the proposed algorithm and debris images which are processed by the proposed algorithm are used to debris recognition. The experimental results show that the algorithm can enhance image texture details automatically, and the image enhancement effects are more obvious. The debris recognition rate has a certain improving effect.
【关闭】
