基于强边缘和稀疏约束的运动模糊图像盲复原

Blind Restoration of Motion Blurred Images Based on Strong Edges and Sparse Constraints

针对图像复原过程中弱边缘会错误指导模糊核的估计以及噪声引起的振铃效应问题,提出一种基于强边缘和稀疏约束的运动模糊图像盲复原算法。首先,基于多尺度的思想,对输入图像进行尺度分解,然后,在每一层尺度上运用双边滤波和冲击滤波对模糊图像进行预处理,通过梯度筛选得到模糊图像的强边缘。求解代价函数中,对模糊图像施加梯度稀疏约束,对点扩散函数施加能量约束,循环迭代得到点扩散函数的最优解。最后再通过非盲解卷积得到最终的复原图像。仿真结果表明,论文算法能有效地复原出图像的轮廓细节。

Aiming at the problem that weak edges will erroneously guide the estimation of fuzzy kernels and the ringing effect caused by noise during image restoration,a blind restoration algorithm for motion blur images based on strong edges and sparse con⁃straints is proposed.First,based on the idea of multi-scale,the input image is scaled.Then,at each level of scale,bilateral filter⁃ing and shock filtering are used to preprocess the blurred image,and the strong edges of the blurred image are obtained by gradient filtering.In solving the cost function,gradient sparse constraints are imposed on the blurred image,and energy constraints are im⁃posed on the point spread function,and the optimal solution of the point spread function is obtained through cyclic iteration.Final⁃ly,the final restored image is obtained by non-blind deconvolution.Simulation results show that the algorithm in this paper can ef⁃fectively restore the outline details of the image.