#ifndef float_image_mmorph_H #define float_image_mmorph_H /* Tools for mathematical morphology operations. */ /* Last edited on 2009-02-16 00:59:07 by stolfi */ #include #include #include #include float_image_t *float_image_mmorph_dilate(float_image_t *A, int hw, double wt[]); /* Computes the dilation of image {A} by the tensor mask {wt'*wt} derived from the unidimensional weights {wt[0..2*hw]}. More precisely, the procedure computes an image {G} with the same size and channel count as {A}, such that each sample {G(c,x,y)} is {max { wt[dx+hw]*wt[dy+hw]*A[c,x+dx,y+dy] : dx,dy \in -hw..+hw }} where {A[c,x,y]} is {float_image_get_sample(A,c,x,y)}. The vector {wt} must be a unidimensional weight table with {2*hw+1} entries. Usually the entries are non-negative and symmetric around {wt[hw]}. */ /* !!! Add {float_image_mmorph_erode} !!! */ #endif