Prototyping object (derived from MIME project)

Functions
cpl_image *	hdrl_get_spatial_freq (cpl_image *ima, double gausfilt, int mirrorx, int mirrory)
	Get low spatial frequency componenets from an image using the FFTW.
cpl_image *	hdrl_mime_image_polynomial_bkg (cpl_image *image, int dim_X, int dim_Y, cpl_matrix **coeffs)
cpl_error_code	hdrl_mime_compute_polynomial_bkg (const cpl_imagelist *images, cpl_imagelist *bkg_images, int dim_X, int dim_Y, cpl_matrix **coeffs)
	Fit smooth background for a list of images.
cpl_matrix *	hdrl_mime_legendre_tensors_create (int nx, int ny, int npx, int npy)
	Create tensor products of Legendre polynomials.
cpl_matrix *	hdrl_mime_matrix_linspace_create (int n, double a, double b)
	Create equally spaced nodes.
cpl_matrix *	hdrl_mime_legendre_polynomials_create (int npoly, double a, double b, const cpl_matrix *x)
	Create the Legendre polynomial basis on the interval (a,b).
cpl_matrix *	hdrl_mime_linalg_pairwise_column_tensor_products_create (const cpl_matrix *mat1, const cpl_matrix *mat2)
	Create selected pairwise tensor products of the columns of two matrices.
cpl_error_code	hdrl_mime_matrix_copy_column (const cpl_matrix *mat1, int j_1, cpl_matrix *mat2, int j_2)
	Copy a column from one matrix to another matrix.
cpl_matrix *	hdrl_mime_linalg_tensor_products_columns_create (const cpl_matrix *mat1, const cpl_matrix *mat2)
	Create the tensor products of the columns of two matrices.
cpl_matrix *	hdrl_mime_tensor_weights_create (int nx, int ny)
	Create tensor product weights.
cpl_error_code	hdrl_mime_matrix_mask_rows (cpl_matrix *mat, const cpl_mask *mask)
	Fill matrix rows with zeros as indicated by a mask.
cpl_error_code	hdrl_mime_matrix_rescale_rows (const cpl_matrix *mat, const cpl_matrix *d, cpl_matrix *dmat)
	Multiply the rows of a matrix by given factors.
cpl_matrix *	hdrl_mime_linalg_solve_tikhonov (const cpl_matrix *mat, const cpl_matrix *rhs, double alpha)
	Solve an overdetermined linear system in the least-squares sense.
cpl_matrix *	hdrl_mime_linalg_normal_equations_create (const cpl_matrix *mat, double alpha)
	Create the matrix transpose(A) * A + alpha for given A and alpha.
cpl_matrix *	hdrl_mime_matrix_product_left_transpose_create (const cpl_matrix *mat1, const cpl_matrix *mat2)
	Create the product of the transpose of a matrix with another matrix.
cpl_error_code	hdrl_mime_matrix_product (const cpl_matrix *mat1, const cpl_matrix *mat2, cpl_matrix *product)
	Fill a matrix with the product of two given matrices.

Detailed Description

This module contains functions derived from MIME project, adapted to HDRL.

Function Documentation

hdrl_get_spatial_freq()

cpl_image * hdrl_get_spatial_freq	(	cpl_image *	ima,
		double	gausfilt,
		int	mirrorx,
		int	mirrory
	)

Get low spatial frequency componenets from an image using the FFTW.

Function to calculate the low spatial frequency

Parameters

ima	image
gausfilt	Gaussian Fourier filter size
mirrorx	for mirroring edges (ocfft continuity)
mirrory	for mirroring edges (ocfft continuity)

Returns

1 newly allocated image.

hdrl_mime_compute_polynomial_bkg()

cpl_error_code hdrl_mime_compute_polynomial_bkg	(	const cpl_imagelist *	images,
		cpl_imagelist *	bkg_images,
		int	dim_X,
		int	dim_Y,
		cpl_matrix **	coeffs
	)

Fit smooth background for a list of images.

Parameters

	images	List of images.
[out]	bkg_images	Smooth background images.
	dim_X,dim_Y	dimensions of polynomials in and .
[out]	coeffs	Polynomial coefficients.

Returns

CPL_ERROR_NONE or the appropriate error code.

This function computes smooth background images by fitting polynomial surfaces to the input images. Bad-pixel masks for the images are taken into account.

hdrl_mime_image_polynomial_bkg()

cpl_image * hdrl_mime_image_polynomial_bkg	(	cpl_image *	image,
		int	dim_X,
		int	dim_Y,
		cpl_matrix **	coeffs
	)

hdrl_mime_legendre_polynomials_create()

cpl_matrix * hdrl_mime_legendre_polynomials_create	(	int	npoly,
		double	a,
		double	b,
		const cpl_matrix *	x
	)

Create the Legendre polynomial basis on the interval (a,b).

Parameters

npoly	Number of polynomials.
a	Left endpoint of the interval.
b	Right endpoint of the interval.
x	Nodes, at which the polynomials are evaluated.

Returns

A matrix with the values of the polynomials at x.

The i-th column contains the values of the i-th polynomial at the given nodes. The polynomials have degrees 0, 1, ..., npoly-1. The nodes must lie on the interval [a, b]. The specific dimensions of the matrix x are not used, only its size.

The returned matrix must be deallocated using cpl_matrix_delete().

hdrl_mime_legendre_tensors_create()

cpl_matrix * hdrl_mime_legendre_tensors_create	(	int	nx,
		int	ny,
		int	npx,
		int	npy
	)

Create tensor products of Legendre polynomials.

Parameters

nx	Number of nodes in the x-direction.
ny	Number of x nodes in the y-direction.
npx	Number of tensor products of functions of x.
npy	Number of tensor products of functions of y.

Returns

The tensor products of Legendre polynomials.

The returned matrix must be deallocated using cpl_matrix_delete().

hdrl_mime_linalg_normal_equations_create()

cpl_matrix * hdrl_mime_linalg_normal_equations_create	(	const cpl_matrix *	mat,
		double	alpha
	)

Create the matrix transpose(A) * A + alpha for given A and alpha.

Parameters

mat	Matrix,
alpha	The regularization parameter.

Returns

The matrix transpose(mat) * mat + alpha.

Note

Only the upper triangle is computed, since cpl_matrix_decomp_chol() only requires the upper triangle.

hdrl_mime_linalg_pairwise_column_tensor_products_create()

cpl_matrix * hdrl_mime_linalg_pairwise_column_tensor_products_create	(	const cpl_matrix *	mat1,
		const cpl_matrix *	mat2
	)

Create selected pairwise tensor products of the columns of two matrices.

Parameters

mat1	A matrix,
mat2	A matrix.

Returns

The tensor product of pairs the columns of the two matrices.

The tensor product of the j1-th and j2-th columns is created iff j1*(nc2-1) + j2*(nc1-1) <= (nc1-1)*(nc2-1). The two matrices may have different dimensions.

hdrl_mime_linalg_solve_tikhonov()

cpl_matrix * hdrl_mime_linalg_solve_tikhonov	(	const cpl_matrix *	mat,
		const cpl_matrix *	rhs,
		double	alpha
	)

Solve an overdetermined linear system in the least-squares sense.

Parameters

mat	A matrix.
rhs	A matrix containing right-hand-side vectors.
alpha	The regularization parameter of the Tikhonov method.

Returns

A matrix with solutions of the least-squares problem.

Typically, this method is used for overdetermined systems, where the matrix has more rows than columns, but it can also be used for square and underdetermined systems. Several right-hand-sides can be provided. The regularization parameter increases with the noise level.

hdrl_mime_linalg_tensor_products_columns_create()

cpl_matrix * hdrl_mime_linalg_tensor_products_columns_create	(	const cpl_matrix *	mat1,
		const cpl_matrix *	mat2
	)

Create the tensor products of the columns of two matrices.

Parameters

mat1	A matrix,
mat2	A matrix.

Returns

The tensor product of the columns of the two matrices.

The two matrices must have the same number of columns. The result has dimensions (nr1*nr2) x nc.

hdrl_mime_matrix_copy_column()

cpl_error_code hdrl_mime_matrix_copy_column	(	const cpl_matrix *	mat1,
		int	j_1,
		cpl_matrix *	mat2,
		int	j_2
	)

Copy a column from one matrix to another matrix.

Parameters

	mat1	The matrix whose column is copied,
	j_1	The index of the column to be copied,
[in,out]	mat2	The matrix whose column is overwritten,
[in,out]	j_2	The index of the column to be overwritten.

Returns

CPL_ERROR_NONE or the appropriate error code.

Both matrices must have the same number of rows.

hdrl_mime_matrix_linspace_create()

cpl_matrix * hdrl_mime_matrix_linspace_create	(	int	n,
		double	a,
		double	b
	)

Create equally spaced nodes.

Parameters

n	The number of nodes,
a	The leftmost node,
b	The rightmost node.

Returns

A column matrix with the equally spaced nodes.

The number of nodes n must be at least 2.

hdrl_mime_matrix_mask_rows()

cpl_error_code hdrl_mime_matrix_mask_rows	(	cpl_matrix *	mat,
		const cpl_mask *	mask
	)

Fill matrix rows with zeros as indicated by a mask.

Parameters

mat	A matrix,
mask	A mask flagging rows to be filled with 0.0s,

Returns

CPL_ERROR_NONE or the appropriate error code.

The size of mask must be equal to the number of rows of mat. The rows corresponding to CPL_BINARY_1 are set to 0.0.

hdrl_mime_matrix_product()

cpl_error_code hdrl_mime_matrix_product	(	const cpl_matrix *	mat1,
		const cpl_matrix *	mat2,
		cpl_matrix *	product
	)

Fill a matrix with the product of two given matrices.

Parameters

	mat1	A matrix,
	mat2	A matrix,
[out]	product	The product of the matrices.

Returns

CPL_ERROR_NONE or the appropriate error code.

The number of rows of mat1 must be equal to the number of rows of product. The number of columns of mat2 must be equal to the number of columns of product. The number of columns of mat1 must be equal to the number of rows of mat2.

hdrl_mime_matrix_product_left_transpose_create()

cpl_matrix * hdrl_mime_matrix_product_left_transpose_create	(	const cpl_matrix *	mat1,
		const cpl_matrix *	mat2
	)

Create the product of the transpose of a matrix with another matrix.

Parameters

mat1	A matrix,
mat2	A matrix.

Returns

The product of the transpose of the first matrix with the second matrix.

The two matrices must have the same number of rows. The product matrix must be deallocated with cpl_matrix_delete().

hdrl_mime_matrix_rescale_rows()

cpl_error_code hdrl_mime_matrix_rescale_rows	(	const cpl_matrix *	mat,
		const cpl_matrix *	d,
		cpl_matrix *	dmat
	)

Multiply the rows of a matrix by given factors.

Parameters

	mat	A matrix,
	d	The factors.
[out]	dmat	The matrix with rescaled rows.

Returns

CPL_ERROR_NONE or the appropriate error code.

The number of rows must be equal to the size of d. The matrix dmat must be allocated before calling this function.

hdrl_mime_tensor_weights_create()

cpl_matrix * hdrl_mime_tensor_weights_create	(	int	nx,
		int	ny
	)

Create tensor product weights.

Parameters

nx	Number of nodes in the x-direction,
ny	Number of x nodes in the y-direction,

Returns

The tensor weights

The returned matrix must be deallocated using cpl_matrix_delete().