Blind multichannel image restoration using oblique projections

We address the problem of image restoration in a multichannel system degraded by unknown blurs and additive noise. Instead of estimating first the channels or the restoration filters, we propose to estimate directly the input image by using a linear estimator in the form of oblique projections (OP) operator. In our OP method, the restoration is achieved by projecting the observed data onto the subspace spanned by the part of image weighted by corner coefficients of the channels while at the same time eliminating the interference part, arising from parts of image weighted by non-corner coefficients, along a different direction. To demonstrate the performance of the proposed method, we run simulaticons and compare the results with those of two similar methods.