Abstract:
To solve the problem that the performance of the non-negativity and support constraint recursive inverse filtering(NAS-RIF) algorithm decreases while it is applied to restore degraded images under a low signal noise ratio(SNR) condition,a NAS-RIF algorithm based on lifting wavelet transform for blind restoration is proposed.Firstly,the degraded image is decomposed by integer lifting wavelet transform,and its wavelet coefficients in wavelet domain are obtained in different frequency sub-bands.Secondly,the improved NAS-RIF algorithm based on space-adaptive and regularization is employed to restore degraded image in each sub-bands.According to different frequency and orientation characteristic in different sub-bands,corresponding regularization operators,regularization parameters and space-weights are adaptively selected to remove blurs in low frequency sub-bands,reduce noise and preserve edges in high frequency sub-bands.Finally,the restored image is reconstructed by the inverse integer lifting wavelet transform.The simulating experiments on two kinds of degraded images are performed under different SNR conditions,and the ΔSNRs by proposed algorithm are 5.849 1 dB and 9.713 6 dB,respectively.The experiment results demonstrate that the proposed algorithm not only obtains better restoration results,but has faster convergence rate.