全局稀疏梯度耦合张量扩散的图像去噪模型

Global sparse gradient coupled tensor diffusion model for image denoising

针对扩散张量和时滞正则模型产生边缘模糊的问题,提出一种全局稀疏梯度耦合张量扩散的图像去噪模型.首先,该模型利用全局稀疏梯度构造张量矩阵;然后,由张量矩阵引导扩散方程去噪,使其在平滑区域为各项同性扩散以去除噪声,在非平滑区域沿切线方向扩散以保护边缘和细节.算法运用显式Euler差分格式求解提出的模型.数值实验表明,无论是客观度量还是视觉效果,文中提出的方法都能得到较好的去噪结果.实验结果说明,利用全局稀疏梯度能得到更加准确和鲁棒的引导图,从而有效提高了原有方法的去噪效果.

A global sparse gradient coupled tensor diffusion model for image denoising is proposed to improve the problem of edges blurring induced by the DTTR model.First,a tensor matrix is constructed by the global sparse gradient which is more accurate and robust than classic gradient operators.Then the diffusion equation is guided by the tensor matrix for image denoising.The diffusion resulting from this model is isotropic inside a homogeneous region and anisotropic along its edge so that an accurate tracking of the edges is possible.Numerical experiments show that the proposed method achieves a competitive denoising performance in comparison with the comparative algorithms in terms of both subjective and objective qualities.Experimental results indicate that the performances of denoising methods can be improved by introducing aguide map,which is obtained by the global sparse gradient model.