Algorithms for Blind Deconvolution

Equation (6.4.20) is not accurate near the boundaries of the subimage, but is a good representation away from the subimage boundaries. From (6.4.20),

	(6.4.21)

By summing both sides of (6.4.21) over all subimages and rewriting the expression, we obtain

	(6.4.22)

Equation (6.4.22) is the basis for estimating . The numerator term is obtained from . The term can be estimated from the empirical observation that

	(6.4.23)

where is obtained from an ungergraded image that is similar in content to . From (6.4.22) and (6.4.23), is estimated by

	(6.4.24)

The performance of the blind deconvlolution method based on (6.4.24) is similar to the method based on (6.4.20)