Contour stencils weighted L1 demosaicing. More...

This graph shows which files directly or indirectly include this file:

Functions
int	CSWL1Demosaic (float *Image, int Width, int Height, int RedX, int RedY, float Alpha, float Epsilon, float Sigma, float Tol, int MaxIter, int ShowEnergy)
	Contour stencils weighted L1 demosaicing. More...

int	DisplayContours (const float Image, int Width, int Height, int RedX, int RedY, const char OutputFile)
	Display orientations estimated by mosaiced contour stencils. More...

void	FitMosaicedStencils (int Stencil, const float Input, int Width, int Height, int RedX, int RedY)
	Estimate the contour orientations of a mosaiced image. More...

Detailed Description

Contour stencils weighted L1 demosaicing.

Author: Pascal Getreuer getre.nosp@m.uer@.nosp@m.gmail.nosp@m..com

This program is free software: you can use, modify and/or redistribute it under the terms of the simplified BSD License. You should have received a copy of this license along this program. If not, see http://www.opensource.org/licenses/bsd-license.html.

Definition in file dmcswl1.h.

Function Documentation

int CSWL1Demosaic	(	float *	Image,
		int	Width,
		int	Height,
		int	RedX,
		int	RedY,
		float	Alpha,
		float	Epsilon,
		float	Sigma,
		float	Tol,
		int	MaxIter,
		int	ShowEnergy
	)

Contour stencils weighted L1 demosaicing.

Parameters

Image	the input RGB image in planar row-major order
Width,Height	the image dimensions
RedX,RedY	the coordinates of the upper-leftmost red pixel
Alpha	weight on the chromatic term
Epsilon	edge weight for weak links in the graph
Sigma	graph filtering parameter
Tol	stopping tolerance
MaxIter	maximum number of iterations
ShowEnergy	if nonzero, display the energy value after each iteration

Returns: 1 on success, 0 on failure

This is the main computation routine for contour stencils demosaicing. It solves the minimization

$\left\{ \begin{aligned} \operatorname*{arg\,min}_{d,u} & \sum_m \Bigl(\sum_n \bigl(w_{m,n} d^L_{m,n} \bigr)^2\Bigr)^{1/2} + \alpha \sum_m \Bigl(\sum_n \Bigl(w_{m,n} \sqrt{(d^{C1}_{m,n})^2 + (d^{C2}_{m,n})^2} \,\Bigr)^2\Bigr)^{1/2} \\ \text{subject to} \; & d_{m,n} = C(u_m - u_n), \; m,n\in\mathbb{Z}^2, \\ & u_m^{k} = f_m, \; m\in\Omega^k, k\in\{R,G,B\}, \end{aligned}\right.$

by Bregman iteration. This is done by alternatingly solving the D-subproblem with DShrink and the U-subproblem with UGaussSeidel.

Definition at line 750 of file dmcswl1.c.

Here is the call graph for this function:

Here is the caller graph for this function:

int DisplayContours	(	const float *	Image,
		int	Width,
		int	Height,
		int	RedX,
		int	RedY,
		const char *	OutputFile
	)

Display orientations estimated by mosaiced contour stencils.

Parameters

Image	the input RGB image in planar row-major order
Width,Height	the image dimensions
RedX,RedY	the coordinates of the upper-leftmost red pixel
OutputFile	EPS file to write

Returns: 1 on success, 0 on failure

Warning: For larger images (e.g., over 300x300), the output EPS file is very large. Some measures are taken for efficient encoding, but compression is not performed.

This routine writes an EPS file of Image superimposed with the orientations detected by mosaiced contour stencils at each pixel. This function is called when running the command line program with the -s option.

For visualization purposes, it is recommended to pass an image that has full color information rather than a mosaiced image (i.e., use the original image or a demosaiced image). The estimated contour orientations are the same regardless of whether the input is mosaiced.

Using the Ghostscript program, the EPS output (say, "contours.eps") can be converted to PDF as

gs -dSAFER -q -P- -dCompatibilityLevel=1.4 -dNOPAUSE -dBATCH \
    -sDEVICE=pdfwrite -sOutputFile=contours.pdf -c .setpdfwrite \
    -f contours.eps

Distiller commands are included within the EPS to preserve the bounding box. PDF conversion applies (lossless) deflate compression to the image data, which significantly decreases the file size.

Definition at line 59 of file displaycontours.c.

void FitMosaicedStencils	(	int *	Stencil,
		const float *	Input,
		int	Width,
		int	Height,
		int	RedX,
		int	RedY
	)

Estimate the contour orientations of a mosaiced image.

Parameters

Stencil	array to store the selected stencils
Input	the mosaiced image
Width,Height	dimensions of the image
RedX,RedY	the coordinates of the upper-leftmost red pixel

Mosaiced contour stencils are applied to estimate the contour orientation at each pixel. The output array Stencil holds the index of the selected stencil at each point. The orientation estimate at the ith pixel is Stencil[i]*M_PI/8 radians:

       5  4  3
     6,   |   ,2
    7  `, | ,`  1
    0-----+-----0
Orientations described
    by Stencil[i]

The same enumeration of orientations is used internally for the TV array.

Definition at line 51 of file mstencils.c.

Here is the caller graph for this function:

Functions

Detailed Description

Function Documentation