改进的自适应非局部均值去噪算法应用与研究  

Application and Research of Improved Adaptive Non-local Mean Algorithm in Image Denoising

作　　者：姚培娟[1] 张志利[1] 张亚娟[1] YAO Peijuan;ZHANG Zhili;ZHANG Yajuan(Department of Engineering,Huanghe S&T University,Zhengzhou 450061)

机构地区：[1]黄河科技学院工学部,郑州450061

出　　处：《计算机与数字工程》2022年第7期1566-1572,共7页Computer & Digital Engineering

基　　金：国家重点研发计划公共安全风险防控与应急技术装备重点专项(编号:SQ2017YFSF060165);河南省民办高等学校品牌专业建设-计算机科学与技术(编号:ZLG201903);2020年度郑州地方高校优秀基层教学组织项目(编号:zzlg202005)资助。

摘　　要：常规非局部均值算法权重计算方式容易受到噪声干扰,使得相似性权重不够准确。为进一步提升非局部均值算法去噪效果,提出一种改进自适应非局部均值算法,应用改进的多尺度小波变换边缘检测算子检测图像边缘,将边缘特征作为约束条件,采用多特征相似性度量方法来衡量图像块的相似性,并自适应的确定滤波参数,最后执行多窗口去噪。仿真结果表明:改进方法的去噪结果图像峰值信噪比平均提高54.2%,平均结构相似性平均值提高至0.82以上。结论认为:相比其他算法,改进方法具有更好的去噪效果,图像边缘纹理等细节信息更清晰,在图像处理方面具有推广应用的前景。The weight calculation method of the conventional non-local mean algorithm is easily interfered by noise,which makes the similarity weight not accurate enough.To further improve the denoising effect of the non-local mean algorithm,an im⁃proved non-local mean algorithm based on edge constraints is proposed.The wavelet transform edge detection operator is used to de⁃tect the edge of the image,and the edge features is used as constraints,a multi-feature similarity measurement method is adopted to measure the similarity of image blocks,and filtering parameters are adaptively determined,and multi-window de-noising is per⁃formed finally.The simulation results show that the de-noising result of the improved method has good visual effect,and the peak signal-to-noise ratio is increased by 54.2%on average,mean structural similarity is increased by 0.82 on average.The conclusion is that compared with other algorithms,the improved method has better de-noising effect,and the detailed information such as im⁃age edge texture is clearer,which has the prospect of application in image processing.

关 键 词：非局部均值算法 高斯噪声 边缘约束 抗噪性能 结构特征 信噪比 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]