一种基于线性化Bregman迭代的图像去模糊新方法

A new algorithm based on linearized Bregman iteration for image deblurring

基于线性化Bregman迭代法带有软阈值算子的A+算法,结合广义逆迭代格式,提出一个新的混乱迭代方法求解图像的去模糊问题。在算法上充分考虑对细节信息的有效利用,以弥补在每步迭代过程中为了去模糊而过滤掉的图像细节特征的损失,达到有效滤波的效果。同时在计算时间和恢复效果之间取得平衡。数值试验结果表明,新方法在提高计算效率的同时还能得到很好的图像恢复效果,特别是细节特征和稀疏纹理的恢复。

A new chaotic iteration for image deblurring was proposed. The algorithm was obtained based on A^＋ linearized Bregman iteration with soft thresholding operator, and combined with generalized inverse iterative formula. Taking full consideration of the effective use of detail information, the algorithm can compensate for the loss of image detail features which is filtered to deblur in each iteration, and achieve effectively filtering. At the same time, a balance between computation time and the recovery effect is considered. The numerical experiments furtherly show that the method can improve computation efficiency and image recovery effect, especially the recovery of the detail features and sparse texture.