contrib/gel/gevd/gevd_float_operators.h File Reference

Operators on float pixels in 2D images. More...

Classes
class	gevd_float_operators

Detailed Description

Operators on float pixels in 2D images.

Operators on pixels in 2D image are divided into following groups:

convolution: * gaussian smoothing, high pass filtering. * 2D/1D separable convolution in x and y, even/odd.
detection: * gradient to find step edges * laplacian to find fold edges * local orientation
pyramid: * shrink/expand by factor of 2 using Burt filters. * wavelet transform
other: * correlation * apply function * projection onto axes
utilities:

float is used throughout to avoid overflow/underflow and conversion on math operations like sqrt, sin, etc... All convolution operators have a running cache of the input, and so the from/to buffers can be the same.

    Modifications
     Van-Duc Nguyen (1990) convolution to support image segmentation
     Van-Duc Nguyen (1992) add pyramid and wavelet
     Van-Duc Nguyen (1996) add circular/reflection at border
     Van-Duc Nguyen (1997) add running sum
     Chuck Stewart  (1997) added SurfaceNormalD, SurfaceCurvatureD, ShrinkBy2_D
     Peter Vanroose July 1999 -made all methods static (class has no data)
     Peter Vanroose Sept 2004 -all methods now support "from" & "to" being equal

Definition in file gevd_float_operators.h.

Classes

Detailed Description