基于梯度和信息熵特性的自适应分数阶微积分图像去噪研究  

Research on Adaptive Fractional Calculus Image Denoising Based on Gradient and Information Entropy

作　　者：康凯[1] KANG Kai(Huazhong Institute of Electro-Optics-Wuhan National Laboratory for Optoelectronics,Wuhan 430223,China)

机构地区：[1]华中光电技术研究所-武汉光电国家研究中心

出　　处：《光学与光电技术》2019年第6期56-65,共10页Optics & Optoelectronic Technology

摘　　要：椒盐噪声常存在于数字图像中,以随机的黑白像素点的形式呈现,降低了图像的处理效率。为去除椒盐噪声,基于梯度和信息熵特性,对自适应分数阶微积分椒盐噪声图像去噪算法进行了研究。该算法中,利用图像的局部特征,对图像的噪声点、边界、纹理区域和平缓的区域进行分割。在分割的基础上,对于不同的像素点,给出关于信息熵和梯度的分数阶的阶次分段函数。实验结果表明,相较于传统去噪算法,提出的自适应分数阶微积分椒盐噪声图像去噪算法能大幅提升PSNR和ENTROPY值,从而在较好地完成去噪的同时,还能抑制图像边界和纹理区域的信息缺失。Salt and pepper noise is a common digital image noise,which makes black and white pixels appear randomly on the image,and has a great impact on various image processing processes.In order to remove salt and pepper noise in image processing,an adaptive fractional calculus algorithm for salt and pepper noise image denoising based on gradient and information entropy is proposed.By combining local structure to segment noise points,edges,texture regions and smooth regions,different fractional orders are related to different pixels and then a segment function related to information entropy and gradient is constructed.The experimental results show that compared with the traditional denoising algorithm,the proposed algorithm can greatly improve the value of the PSNR and ENTROPY.Consequently,the proposed adaptive fractional calculus salt and pepper noise denoising algorithm can effectively overcome the shortcoming of image detail information loss,while suppressing salt and pepper noise,better preserving the image details texture and boundary information.

关 键 词：梯度 信息熵 自适应 分数阶 图像去噪 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]