High-level functions to compute detector features

Functions
cpl_error_code	cpl_detector_interpolate_rejected (cpl_image *self)
	Interpolate any bad pixels in an image and delete the bad pixel map.
cpl_error_code	cpl_flux_get_bias_window (const cpl_image *bias_image, const cpl_size *zone_def, cpl_size ron_hsize, cpl_size ron_nsamp, double *bias, double *error)
	Compute the bias in a rectangle.
cpl_error_code	cpl_flux_get_noise_ring (const cpl_image *diff, const double *zone_def, cpl_size ron_hsize, cpl_size ron_nsamp, double *noise, double *error)
	Compute the readout noise in a ring.
cpl_error_code	cpl_flux_get_noise_window (const cpl_image *diff, const cpl_size *zone_def, cpl_size ron_hsize, cpl_size ron_nsamp, double *noise, double *error)
	Compute the readout noise in a rectangle.

Detailed Description

Synopsis:

#include "cpl_detector.h"

Function Documentation

cpl_detector_interpolate_rejected()

cpl_error_code cpl_detector_interpolate_rejected

(

cpl_image *

self

)

Interpolate any bad pixels in an image and delete the bad pixel map.

Parameters

self	The image to clean

Returns

The _cpl_error_code_ or CPL_ERROR_NONE

The value of a bad pixel is interpolated from the good pixels among the 8 nearest. (If all but one of the eight neighboring pixels are bad, the interpolation becomes a nearest neighbor interpolation). For integer images the interpolation in done with floating-point and rounded to the nearest integer.

If there are pixels for which all of the eight neighboring pixels are bad, a subsequent interpolation pass is done, where the already interpolated pixels are included as source for the interpolation.

The interpolation passes are repeated until all bad pixels have been interpolated. In the worst case, all pixels will be interpolated from a single good pixel.

Possible _cpl_error_code_ set in this function:

• CPL_ERROR_NULL_INPUT if an input pointer is NULL
• CPL_ERROR_DATA_NOT_FOUND if all pixels are bad

References cpl_ensure_code, cpl_error_get_code(), CPL_ERROR_INVALID_TYPE, CPL_ERROR_NONE, CPL_ERROR_NULL_INPUT, cpl_image_accept_all(), cpl_image_get_bpm_const(), cpl_image_get_data(), cpl_image_get_size_x(), cpl_image_get_size_y(), cpl_image_get_type(), cpl_mask_get_data_const(), CPL_TYPE_DOUBLE, CPL_TYPE_DOUBLE_COMPLEX, CPL_TYPE_FLOAT, CPL_TYPE_FLOAT_COMPLEX, and CPL_TYPE_INT.

cpl_flux_get_bias_window()

cpl_error_code cpl_flux_get_bias_window	(	const cpl_image *	bias_image,
		const cpl_size *	zone_def,
		cpl_size	ron_hsize,
		cpl_size	ron_nsamp,
		double *	bias,
		double *	error
	)

Compute the bias in a rectangle.

Parameters

bias_image	Input bias image
zone_def	Zone where the bias is to be computed.
ron_hsize	to specify half size of squares (<0 to use default)
ron_nsamp	to specify the nb of samples (<0 to use default)
bias	Output parameter: bias in the frame.
error	Output parameter: error on the bias.

Returns

CPL_ERROR_NONE on success or the relevant _cpl_error_code_ on error

See also

cpl_flux_get_noise_window(), rand()

Note

No calls to srand() are made from CPL

References CPL_ERROR_NONE.

cpl_flux_get_noise_ring()

cpl_error_code cpl_flux_get_noise_ring	(	const cpl_image *	diff,
		const double *	zone_def,
		cpl_size	ron_hsize,
		cpl_size	ron_nsamp,
		double *	noise,
		double *	error
	)

Compute the readout noise in a ring.

Parameters

diff	Input image, usually a difference frame.
zone_def	Zone where the readout noise is to be computed.
ron_hsize	to specify half size of squares (<0 to use default)
ron_nsamp	to specify the nb of samples (<0 to use default)
noise	On success, *noise is the noise in the frame.
error	NULL, or on success, *error is the error in the noise

Returns

CPL_ERROR_NONE on success or the relevant _cpl_error_code_ on error

See also

cpl_flux_get_noise_window(), rand()

Note

No calls to srand() are made from CPL

The provided zone is an array of four integers specifying the zone to take into account for the computation. The integers specify a ring as x, y, r1, r2 where these coordinates are given in the FITS notation (x from 1 to nx, y from 1 to ny and bottom to top).

Possible _cpl_error_code_ set in this function:

• CPL_ERROR_NULL_INPUT if diff noise or zone_def is NULL
• CPL_ERROR_ILLEGAL_INPUT if the internal radius is bigger than the external one in zone_def
• CPL_ERROR_DATA_NOT_FOUND if an insufficient number of samples were found inside the ring

References cpl_bivector_delete(), cpl_bivector_get_x_data_const(), cpl_bivector_get_y_data_const(), cpl_ensure_code, CPL_ERROR_DATA_NOT_FOUND, CPL_ERROR_ILLEGAL_INPUT, CPL_ERROR_NONE, CPL_ERROR_NULL_INPUT, cpl_errorstate_get(), cpl_errorstate_is_equal(), cpl_errorstate_set(), cpl_image_get_size_x(), cpl_image_get_size_y(), cpl_image_get_stdev_window(), CPL_SIZE_FORMAT, cpl_vector_delete(), cpl_vector_get_median(), cpl_vector_get_stdev(), cpl_vector_new(), cpl_vector_set(), cpl_vector_unwrap(), and cpl_vector_wrap().

cpl_flux_get_noise_window()

cpl_error_code cpl_flux_get_noise_window	(	const cpl_image *	diff,
		const cpl_size *	zone_def,
		cpl_size	ron_hsize,
		cpl_size	ron_nsamp,
		double *	noise,
		double *	error
	)

Compute the readout noise in a rectangle.

Parameters

diff	Input image, usually a difference frame.
zone_def	Zone where the readout noise is to be computed.
ron_hsize	to specify half size of squares (<0 to use default)
ron_nsamp	to specify the nb of samples (<0 to use default)
noise	Output parameter: noise in the frame.
error	Output parameter: error on the noise.

Returns

CPL_ERROR_NONE on success or the relevant _cpl_error_code_ on error

See also

rand()

Note

No calls to srand() are made from CPL

This function is meant to compute the readout noise in a frame by means of a MonteCarlo approach. The input is a frame, usually a difference between two frames taken with the same settings for the acquisition system, although no check is done on that, it is up to the caller to feed in the right kind of frame.

The provided zone is an array of four integers specifying the zone to take into account for the computation. The integers specify ranges as xmin, xmax, ymin, ymax, where these coordinates are given in the FITS notation (x from 1 to lx, y from 1 to ly and bottom to top). Specify NULL instead of an array of four values to use the whole frame in the computation.

The algorithm will create typically 100 9x9 windows on the frame, scattered optimally using a Poisson law. In each window, the standard deviation of all pixels in the window is computed and this value is stored. The readout noise is the median of all computed standard deviations, and the error is the standard deviation of the standard deviations.

Both noise and error are returned by modifying a passed double. If you do not care about the error, pass NULL.

Possible _cpl_error_code_ set in this function:

• CPL_ERROR_NULL_INPUT if diff or noise is NULL
• CPL_ERROR_ILLEGAL_INPUT if the specified window (zone_def) is invalid

References CPL_ERROR_NONE.