pderiche

Computes gradient magnitude and maxima localization using Deriche algorithm.

Synopsis

pderiche sigma [-m mask] [im_in|-] [im_mag|-] [im_dir|-]

Description

pderiche computes the gradient magnitude and the gradient direction images and performs the maxima localization. The output image can then be used to locate the contours. The output image im_out is built with the maximum magnitude value in the direction of the gradient. Other values are set to 0.

The gradient magnitude value reflects the amount of grayscale variation in this point. The more is the variation, the greater is the value.

The gradient extraction and localization is done in three steps :

smoothing,
gradient computing,
local maxima extraction.

The direction is the atan(dy/dx) measured in radians. The direction image im_dir is necessarily of type float (values in [0..2*PI]).
Warning: the direction follows the image coordinate system, it means that it's going clockwise when displayed, since the y coordinates are inversed in the image system.

Note: The image border of size 1 is set to 0.

Parameters

sigma is a real value that controls the strength of the smoothing. Values are generally from the interval [0..10]. The lower is the sigma value, the stronger is the smoothing and thus the less there are contour points in the output image.

Inputs

im_in: a 2D grayscale image.

Outputs

im_out: the gradient magnitude image of same type as the input image.
im_dir: the gradient direction image of type float.

Result

Returns SUCCESS or FAILURE.

Examples

Performs an edge detection for the tangram.pan image:

   pderiche 1 tangram.pan a.pan b.pan
   pbinarization 10 1e30 a.pan c.pan

C++ prototype

Errc PDeriche( const Img2duc &im_in, Img2duc &im_mag, Img2duc &im_dir, float sigma );

Version française

Détection et localisation des contours par l'algorithme de Deriche.

Author: Carlotti & Joguet