基于双边滤波的各向异性扩散方程图像去噪  

Anisotropic Diffusion Image Denoising based on Bilateral Filtering

作　　者：许冠军[1] 喻晓[2] 

机构地区：[1]台州科技职业学院信息学院,浙江台州318020 [2]上饶师范学院数学与计算机科学学院,江西上饶334001

出　　处：《上饶师范学院学报》2017年第3期11-16,共6页Journal of Shangrao Normal University

基　　金：国家自然科学基金(11561057);江西省自然科学基金(20151BAB211002);浙江省教育技术研究规划课题(JA006);江西省教育厅科技项目(GJJ151054)

摘　　要：近年来,各向异性扩散方程模型在图像去噪和特征提取中得到广泛应用。双边滤波为非迭代图像去噪算法,此方法因简单而高效而备受关注。在对两种模型优缺点分析的基础上,结合双边滤波的非线性特点,用双边滤波预光滑,在各向异性扩散方程中使用预光滑图像进行边界检测。和Alvarez等使用高斯核卷积提取边界相比,双边滤波体现了预处理的各向异性,提高了边界检测的准确性。改进的扩散方程的强迫项,在保证去噪效果的同时,提高了迭代的收敛速度。In recent decades,anisotropic diffusion model was widely used in image denoising and edge detection.Bilateral filtering image denoising is highly concerned because of it＇s non-iterative,simple and efficient. In this paper both advantages and disadvantages of the two algorithms are studied. Based on the characteristics of bilateral filtering and anisotropic diffusion, using the bilateral filtering to substitute Gaussian convolution in Alvarez model , which improved edge detection. The forcing term is introduced to ensure the denoising performance, and improves the convergence rate of the iteration.

关 键 词：双边滤波 各向异性扩散方程 扩散系数 边界检测 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]