CE GNC - Contrast Enhancement - Graduated Non Convex restoration

(12/27/93)

PURPOSE

Applies graduated non-convex restoration algorithm to an image.   Example.

SEE ALSO

CE FIT [Contrast Enhancement - FIT the histogram]
CE MED [Contrast Enhancement - Median Filtering]

USAGE

.OPERATION: CE GNC

.INPUT FILE: PIC001
[Enter name of picture to be processed.]

.OUTPUT FILE: PIC002
[Enter name of file receiving the output picture.]

.LAMBDA: 3
[The parameter LAMBDA is a characteristic length or 'scale'. The lower LAMBDA, the finer the structure that is found.]

.H0: 0.02
[The ratio H0 [=sqrt(2*alpha/LAMBDA)] is a 'contrast' sensitivity threshold determining the minimum contrast for detection of an isolated step edge. A step edge in the data counts as isolated if there are no features within a distance LAMBDA of it.]

.EPS: 1.0E-8
[EPS indicates the accuracy of restoration. The smaller EPS, the longer computation time. Reasonable results can be obtained for EPS<=1.0E-7]

NOTES

  1. The ratio g = H0/(2*LAMBDA) is a limit on the gradient above which spurious discontinuities may be generated. If the gradient exceeds g, one or mores discontinuities may appear in the fitted function.

  2. The parameter alpha is a measure of immunity to noise. If the mean noise has standard dev. sigma, then no spurious discontinuities are generated provided alpha>2*sigma**2, approximately.

  3. This program is highly recommended for restoration of noisy pictures. It applies a graduated non-convex algorithm to find the solution of the weak continuity constraints problem for a given picture. Weak continuity constraints prefer continuity, but allow occasional discontinuities if that makes for a simpler overall description. For a detailed discussion of the method and parameters values look in Visual Reconstruction, Andrew Blake & Andrew Zisserman.

  4. Implemented by: Paul Penczek.

SUBROUTINES

GNC, GNC2S, GP, ERC

CALLER

UTIL2

© Copyright Notice /       Enquiries: spider@wadsworth.org