摘要
常规非局部均值算法易受噪声对图像的自相似度计算精度的影响,去噪结果对原始图像的边缘细节信息损伤较多。采用改进的Facet算子提取图像的边缘特征,根据图像内部像素分布情况,在不同的区域采用不同的自相似度计算方法,设置一种变尺寸的搜索窗口,最大限度地搜寻相似性邻域,降低噪声对自相似度计算精度的影响,有效保持图像边缘信息。数据测试结果表明,改进的非局部均值滤波算法能够有效保持边缘纹理信息,去噪效果要优于常规非局部均值滤波算法。
The conventional non local mean filtering algorithm is easy to be affected by the noise on the accuracy of the image self-similarity calculation,and the denoising results are easy to damage the edge details of original images.The improved facet operator was used to extract the edge features of the image.According to the distribution of pixels in the image,different self-similarity calculation methods were used in different region,and a variable size search window was set up,to search similarity neighborhood to the maximum extent and reduce the impact of noise on the accuracy of self-similarity calculation,and effectively maintain the edge of the image information.The data test results show that the improved non local mean filtering algorithm can effectively keep the edge texture information,and its denoising effect is better than the conventional non local mean filtering algorithm.
作者
齐德明
Qi Deming(School of Medical Imaging,Qilu Medical University,Zibo 255300,Shandong,China)
出处
《计算机应用与软件》
北大核心
2021年第9期256-261,279,共7页
Computer Applications and Software
基金
山东省职业教育教学改革研究项目(2017152)。
