Fine granularity spatially adaptive regularization for TVL1 based image deblurring

A total variation weighted l₁ (TVWL1) norm based image dublurring algorithm is proposed. The proposed algorithm uses a series of data matrix to weight the error signal and then the l₁ norms of the resultant series of error signals are used to produce the fidelity term while the regularization term remains the conventional total variation regularization. An alternate minimization approach is used to solve the minimization problem that comprises the fidelity and the regularization terms. It is shown through simulation results that the proposed TVWL1 algorithm offers the same robustness with respect to impulsive noise as that achieved by using the recently proposed total variation l₁ (TVL1) algorithm, while yielding an improved signal-to-noise ratio (SNR), and hence, improved restoration in image deblurring.