Efficient Deconvolution Algorithm

This thesis describes the design of a new deconvolution approach known as Efficient Deconvolution algorithm so that reconstruction image will be free from both blur and noise. This algorithm comprises two steps, a global blur compensation using generalized Wiener filter, followed by a pure local denoising algorithm using dual-tree complex wavelet transform (DT-CWT). The main advantage of this decomposition is that the deconvolution problem is converted into an easier nonwhite noise removal problem, at the likely price of losing global optimality. Given the increasingly high performance of current denoising methods, this approach has become an appealing practical alternative. Our algorithm has also exploited the properties of DT-CWT viz. nearly shift invariant, excellent directionality and explicit phase information to remove noise without the need for assuming a specific noise model. This algorithm is applicable to all ill-conditioned problems, unlike the purely wavelet based deconvolution algorithm WVD; moreover, its estimate features minimal ringing, unlike the purely Fourier based Wiener deconvolution. Also, the performance of this system is demonstrated based on some comparisons with the best available wavelet-based image denoising techniques.