Efficient discrete spatial techniques for blur support identification in blind image deconvolution

This paper proposes two discrete spatial techniques for identifying blur support in blind image deconvolution. Blur identification is a challenging problem in blind image deconvolution. In particular, if the blur support can be estimated reliably in the beginning of restoration, the computational cost of many blind deconvolution schemes can be reduced significantly, and their convergence performance improved. This paper proposes two methods called maximum average square difference (MASD) and maximum average absolute difference (MAAD). They are derived from the autoregressive (AR) model of the underlying images. The efficiency and validity of the techniques are also analyzed in this paper. A main advantage of the proposed techniques is their algorithmic and implementation simplicity. Experimental results show that they are effective in identifying the blur support reliably, thus providing a sound foundation for further blind image deconvolution.