Pink 0.9
|
directional filter for curvilinear feature extraction More...
directional filter for curvilinear feature extraction
Usage: directionalfilter.c in.pgm width length ndir out.pgm
Description: Let F be the original image from in.pgm . This operator computes the supremum of the convolutions of F by a series of kernels K0, ... Kn where n = ndir - 1, which are defined by, for each (x,y) and each i in [0...n]:
sigma = 1 / (2*width*width); lambda = 1 / (2*length*length); theta = i * PI / n; xr = cos(theta) * x - sin(theta) * y; yr = sin(theta) * x + cos(theta) * y; Ki(x,y) = exp(-lambda*yr*yr) * (4*sigma*sigma*xr*xr - 2*sigma) * exp(-sigma*xr*xr)
Types supported: byte 2d, int32_t 2d, float 2d
Category: signal