|
The
image out of focus is thought to be deformed from original image by some
linear transformation. This transformation is easily expressed in the form of
matrix based on the geometrical optics. The problem is to get its inverse
transformation. Usual convolution method for sharpening gives the approximate
solution as iteration method for linear equation. However the iteration by
this method doesn't have convergence because the matrix representing the transformation
doesn't satisfy the convergence conditions (i.e. superior diagonal, etc.) The
method used in this software can get convergence without satisfying the above
conditions as dispersing the variance among the points surrounding the center
point as illustrated in previous webpage. Namely it can make every
eigenvalues less than 1.
(For further details, please visit New Iteration
Method.)
|
