C:/Users/Malcolm/Documents/temp/binaryimagelibrary/canny.c File Reference

#include <stdlib.h>
#include <string.h>
#include <math.h>

Data Structures
struct	CANNY

Macros
#define	ffabs(x)   ( (x) >= 0 ? (x) : -(x) )
#define	GAUSSIAN_CUT_OFF   0.005f
#define	MAGNITUDE_SCALE   100.0f
#define	MAGNITUDE_LIMIT   1000.0f
#define	MAGNITUDE_MAX   ((int) (MAGNITUDE_SCALE * MAGNITUDE_LIMIT))

Functions
unsigned char *	canny (unsigned char *grey, int width, int height)
unsigned char *	cannyparam (unsigned char *grey, int width, int height, float lowThreshold, float highthreshold, float gaussiankernelradius, int gaussiankernelwidth, int contrastnormalised)

Detailed Description

The code detects edges in greyscale images. The method was developed by John F Canny and is generally considered to be the best edge detection method currently available.

C version. Based on Java code by Tom Gibara

Function Documentation

unsigned char * canny	(	unsigned char *	grey,
		int	width,
		int	height
	)

Canny edge detection with default parameters

Parameters

grey	- the greyscale image
width	- image width
height	- image height

Returns

Binary image with edges as set pixels.

Margaret Thatcher (1925-2013) greyscale photograph

Mrs Thatcher edge detected

unsigned char * cannyparam	(	unsigned char *	grey,
		int	width,
		int	height,
		float	lowthreshold,
		float	highthreshold,
		float	gaussiankernelradius,
		int	gaussiankernelwidth,
		int	contrastnormalised
	)

Canny edge detection with parameters passed in by user.

Parameters

grey	- the greyscale image
width	- image width
height	- image height
lowthreshold	- default 2.5
highthreshold	- default 7.5
gaussiankernelradius	- radius of edge detection Gaussian, in standard deviations (default 2.0)
gaussiankernelwidth	- width of Gaussian kernel, in pixels (default 16)
contrastnormalised	- flag to normalise image before edge detection (defualt 0)

Returns

: Binary image with set pixels as edges.