Improved image restoration using wavelet-based denoising and fourier-based deconvolution

Deconvolution of images is an ill-posed problem, which is very often tackled by using the diagonalization property of the circulant matrix in the discrete Fourier transform (DFT) domain. On the other hand, the discrete wavelet transform (DWT) has shown significant success for image denoising because of its space-frequency localization. In this paper, we propose an iterative image restoration algorithm, wherein the DFT-based adaptive regularized constraint total least-squares deconvolution is performed followed by our previously proposed DWT-based maximum a posteriori estimator. The convergence of the proposed method is assured. Experimental results on standard images show that the proposed method provides a restoration performance, which is better than that of several existing methods in terms of signal-to-noise ratio and visual quality.