基于联合规整化约束的图像盲复原

在图像复原研究中加入合适的规整化约束能够有效地提高复原效果。目前普遍将空域上的梯度稀疏性或小波域上的系数稀疏性作为规整化项,但二者并未同时使用。本文提出了对图像采用一种新的空域局部阈值处理和小波域处理的联合规整化复原算法。本算法结合了二者的优点,能够更加充分地利用图像特征信息。实验结果表明,本方法得到的结果有较好的图像质量和较高的算法鲁棒性,能更有效地抑制噪声和伪迹。

Blind deconvolution based on fusion of regularization terms

Adopting suitable regularizations could obtain better deblurring results from blurred images. The sparseness of spatial gradients and wavelet coefficients are mainly used as regulation terms, the two regulation terms are not simultaneously used. A method based on the fusion of spatial regularization term and wavelet regularization term is proposed. For the spatial regularization term, new local threshold is used. The method combines advantages of the two regularization terms and can use more feature in- formation. Experiments indicate that the proposed method is more robust and can recover images with better removal of noise and artifacts.