Image Operations

Collaboration diagram for Image Operations:

Files
file	ops.h

Macros
#define	ICV_DIFF_PHASH   0
#define	ICV_DIFF_SSIM   1

Enumerations
enum	ICV_RESIZE_METHOD { ICV_RESIZE_UNDERSAMPLE , ICV_RESIZE_SHRINK , ICV_RESIZE_NINTERP , ICV_RESIZE_BINTERP }

Functions
int	icv_sanitize (icv_image_t *img)
int	icv_add_val (icv_image_t *img, double val)
int	icv_multiply_val (icv_image_t *img, double val)
int	icv_divide_val (icv_image_t *img, double val)
int	icv_pow_val (icv_image_t *img, double val)
icv_image_t *	icv_add (icv_image_t *img1, icv_image_t *img2)
icv_image_t *	icv_sub (icv_image_t *img1, icv_image_t *img2)
icv_image_t *	icv_multiply (icv_image_t *img1, icv_image_t *img2)
icv_image_t *	icv_divide (icv_image_t *img1, icv_image_t *img2)
int	icv_saturate (icv_image_t *img, double sat)
int	icv_resize (icv_image_t *bif, ICV_RESIZE_METHOD method, size_t out_width, size_t out_height, size_t factor)
int	icv_rot (size_t argc, const char *argv[])
int	icv_diff (int *matching, int *off_by_1, int *off_by_many, icv_image_t *img1, icv_image_t *img2)
icv_image_t *	icv_diffimg (icv_image_t *img1, icv_image_t *img2)
DEPRECATED uint32_t	icv_pdiff (icv_image_t *img1, icv_image_t *img2)
fastf_t	icv_adiff (icv_image_t *img1, icv_image_t *img2, int m)
int	icv_diff_render_info (const icv_image_t *img1, const icv_image_t *img2, struct bu_vls *out_msgs)
int	icv_diff_nirt_shots (const icv_image_t *img1, const icv_image_t *img2, FILE *nirt_out1, FILE *nirt_out2)
int	icv_fit (icv_image_t *img, struct bu_vls *msg, size_t o_width_req, size_t o_height_req, fastf_t sf)

Detailed Description

Various routines to perform operations on images.

Macro Definition Documentation

ICV_DIFF_PHASH

#define ICV_DIFF_PHASH   0

Compare two images using approximate techniques.

Definition at line 237 of file ops.h.

ICV_DIFF_SSIM

#define ICV_DIFF_SSIM   1

Definition at line 238 of file ops.h.

Enumeration Type Documentation

ICV_RESIZE_METHOD

enum ICV_RESIZE_METHOD

Enumerator
ICV_RESIZE_UNDERSAMPLE
ICV_RESIZE_SHRINK
ICV_RESIZE_NINTERP
ICV_RESIZE_BINTERP

Definition at line 147 of file ops.h.

Function Documentation

icv_sanitize()

int icv_sanitize

(

icv_image_t *

img

)

This function sanitizes the image.

It forces the image pixels to be in the prescribed range.

All the pixels higher than the max range are set to MAX (1.0). All the pixels lower than the min range are set to MIN (0.0).

Note if an image(bif) is sanitized then, (bif->flags&&ICV_SANITIZED) is true.

icv_add_val()

int icv_add_val	(	icv_image_t *	img,
		double	val
	)

This adds a constant value to all the pixels of the image. Also if the flag ICV_OPERATIONS_MODE is set this doesn't sanitize the image.

Note to set the flag for a bif (icv_image struct); bif->flags |= ICV_OPERATIONS_MODE;

icv_multiply_val()

int icv_multiply_val	(	icv_image_t *	img,
		double	val
	)

This multiplies all the pixels of the image with a constant Value. Also if the flag ICV_OPERATIONS_MODE is set this doesn't sanitize the image.

icv_divide_val()

int icv_divide_val	(	icv_image_t *	img,
		double	val
	)

This divides all the pixels of the image with a constant Value. Also if the flag ICV_OPERATIONS_MODE is set this doesn't sanitize the image.

icv_pow_val()

int icv_pow_val	(	icv_image_t *	img,
		double	val
	)

This raises all the pixels of the image to a constant exponential power. Also if the flag ICV_OPERATIONS_MODE is set this doesn't sanitize the image.

icv_add()

icv_image_t * icv_add	(	icv_image_t *	img1,
		icv_image_t *	img2
	)

This routine adds pixel value of one image to pixel value of other pixel and inserts in the same index of the output image.

Also it sanitizes the image.

icv_sub()

icv_image_t * icv_sub	(	icv_image_t *	img1,
		icv_image_t *	img2
	)

This routine subtracts pixel value of one image from pixel value of other pixel and inserts the result at the same index of the output image.

Also it sanitizes the image.

Parameters

img1	First Image.
img2	Second Image.

Returns

New icv_image (img1 - img2)

icv_multiply()

icv_image_t * icv_multiply	(	icv_image_t *	img1,
		icv_image_t *	img2
	)

This routine multiplies pixel value of one image to pixel value of other pixel and inserts the result at the same index of the output image.

Also it sanitizes the image.

Parameters

img1	First Image.
img2	Second Image.

Returns

New icv_image (img1 * img2)

icv_divide()

icv_image_t * icv_divide	(	icv_image_t *	img1,
		icv_image_t *	img2
	)

This routine divides pixel value of one image from pixel value of other pixel and inserts the result at the same index of the output image.

Also it sanitizes the image.

Parameters

img1	First Image.
img2	Second Image.

Returns

New icv_image (img1 / img2)

icv_saturate()

int icv_saturate	(	icv_image_t *	img,
		double	sat
	)

Change the saturation of image pixels. If sat is set to 0.0 the result will be monochromatic; if sat is made 1.0, the color will not change; if sat is made greater than 1.0, the amount of color is increased.

Parameters

img	RGB Image to be saturated.
sat	Saturation value.

icv_resize()

int icv_resize	(	icv_image_t *	bif,
		ICV_RESIZE_METHOD	method,
		size_t	out_width,
		size_t	out_height,
		size_t	factor
	)

This function resizes the given input image. Mode of usage: a) ICV_RESIZE_UNDERSAMPLE : This method undersamples the said image e.g. icv_resize(bif, ICV_RESIZE_UNDERSAMPLE, 0, 0, 2); undersamples the image with a factor of 2.

b) ICV_RESIZE_SHRINK : This Shrinks the image, keeping the light energy per square area as constant. e.g. icv_resize(bif, ICV_RESIZE_SHRINK,0,0,2); shrinks the image with a factor of 2.

c) ICV_RESIZE_NINTERP : This interpolates using nearest neighbor method. Output endpoints map to input endpoints; a one-pixel output maps to source coordinate zero. e.g. icv_resize(bif, ICV_RESIZE_NINTERP,1024,1024,0); interpolates the output image to have the size of 1024X1024.

d) ICV_RESIZE_BINTERP : This interpolates using bilinear Interpolation Method. Output endpoints map to input endpoints; a one-pixel output maps to source coordinate zero. Stretching one dimension while shrinking the other is supported. e.g. icv_resize(bif, ICV_RESIZE_BINTERP,1024,1024,0); interpolates the output image to have the size of 1024X1024.

resizes the image inplace.

Parameters

bif	Image (packed in icv_image struct)
method	One of the modes.
out_width	Out Width.
out_height	Out Height.
factor	Integer type data representing the factor to be shrunken

Returns

0 on success and -1 on failure.

icv_rot()

int icv_rot ( size_t  argc, const char *  argv[]  )

extern

Rotate an image. s [-rifb | -a angle] [-# bytes] [-s squaresize] [-w width] [-n height] [-o outputfile] inputfile [> outputfile]

icv_diff()

int icv_diff ( int *  matching, int *  off_by_1, int *  off_by_many, icv_image_t *  img1, icv_image_t *  img2  )

extern

Compare two images and report pixel differences. Return code is 1 if there are any differences, else 0. For more detailed reporting, pass non-null integer pointers to the matching, off_by_1, and/or off_by_many parameters.

Counts are per-channel (matching pixdiff "bytes" semantics):

• matching = number of channel bytes with identical values across both images
• off_by_1 = number of channel bytes differing by exactly 1
• off_by_many = number of channel bytes differing by more than 1

icv_diffimg()

icv_image_t * icv_diffimg ( icv_image_t *  img1, icv_image_t *  img2  )

extern

Generate a visual representation of the differences between two images. (At least for now, images must be the same size.)

For pixels that match, a half-intensity greyscale representation is output. For pixels that differ, each channel is independently highlighted:

• channel byte values differ by exactly 1 → 0xC0 for that channel
• channel byte values differ by more than 1 → 0xFF for that channel
• channel byte values are equal → 0x00 for that channel

This means a pixel that differs only in the red channel appears as a pure red highlight, green-only difference as pure green, etc. – matching the traditional pixdiff output convention.

Returns NULL if there is an error.

icv_pdiff()

DEPRECATED uint32_t icv_pdiff ( icv_image_t *  img1, icv_image_t *  img2  )

extern

Compare two images using perceptual image hashing and report the Hamming distance between them. Useful for approximate image comparisons.

icv_adiff()

fastf_t icv_adiff ( icv_image_t *  img1, icv_image_t *  img2, int  m  )

extern

icv_diff_render_info()

int icv_diff_render_info ( const icv_image_t *  img1, const icv_image_t *  img2, struct bu_vls *  out_msgs  )

extern

Compare the embedded render metadata (icv_render_info) of two images.

Writes a human-readable report to out_msgs describing any differences found in the db filename, object list, and camera parameters.

Returns 0 if the metadata in both images is identical (or both absent), 1 if they differ, and -1 if neither image carries any metadata.

Parameters

img1	First image (may be NULL – treated as having no metadata)
img2	Second image (may be NULL – treated as having no metadata)
out_msgs	bu_vls to receive the human-readable comparison report; may be NULL

icv_diff_nirt_shots()

int icv_diff_nirt_shots ( const icv_image_t *  img1, const icv_image_t *  img2, FILE *  nirt_out1, FILE *  nirt_out2  )

extern

Generate nirt shotline commands for every pixel that differs between img1 and img2.

A separate nirt script is written for each image that has render_info attached. Both scripts encode shotlines for the same set of differing pixels; they differ only in the camera parameters used to reconstruct the rays and in the scene header comment. This allows the caller to interrogate either scene independently even when the images were rendered from different .g files or with different object sets.

Ray reconstruction mirrors BRL-CAD rt/grid.c grid_setup() for the orthographic case (rt_perspective == 0). Perspective is also handled.

Either output file pointer may be NULL to suppress that output. If an image has no render_info the corresponding output is silently skipped.

Parameters

img1	First image
img2	Second image
nirt_out1	Open FILE* to write img1's nirt script (may be NULL)
nirt_out2	Open FILE* to write img2's nirt script (may be NULL)

Returns the number of differing pixels for which shots were written, or -1 on error (e.g., neither active output has render metadata, mismatched sizes).

icv_fit()

int icv_fit ( icv_image_t *  img, struct bu_vls *  msg, size_t  o_width_req, size_t  o_height_req, fastf_t  sf  )

extern

Fit an image to suggested dimensions.