dimage.combine package¶

Notes and to-do:¶

• Does not adjust for pixel area changes; keeps image in whatever units it was input in.
• Uses simple TAN projection and square image; doesn’t yet take arbitrary WCS inputs.
• Will fail on images which span > 180 deg in RA, due to simplistic handling of 0/360 boundary

dimage.combine.combine_apodize(nx, ny, edge=10)[source]¶

Create apodized weight image

Parameters:	nx, ny : int dimensions of image edge : int pixel width of apodized edge (default 10)
Returns:	weight : np.float32 [nx, ny] ndarray weighted image

Notes

Uses a cos() function transition from zero to unity in each dimension along each edge.

dimage.combine.combine_wise(ra, dec, size, band, kernel='lanczos', dampsinc=2.47, lanczos=2.0, edge=10, pixscale=2.75, unmasked=False)[source]¶

Combine WISE images onto a desired footprint

Parameters:	ra, dec : fp.float64 center of image (J2000 deg) size : float size of square image (deg) band : str band name (‘W1’, ‘W2’, ‘W3’, ‘W4’) edge : int pixel width of apodized edge to use (default 10) kernel, dampsinc, lanczos : parameters for dresample pixscale : float output pixel scale in arcsec/pixel (default 2.75) unmasked : bool if True, use unmasked coadds (default False)
Returns:	(image, ivar, wcs) : (np.float32, np.float32, WCS) [nx, ny] image [nx, ny] image inverse variance WCS object

dimage.combine.dcombine(ra, dec, size, pixscale, image_files, ivar_files=None, kernel='lanczos', dampsinc=2.47, lanczos=2.0, edge=10)[source]¶

Combine a list of images onto a desired footprint

Parameters:	ra, dec : np.float64 center of image (J2000 deg) size : float size of square image (deg) pixelscale : float pixel scale (arcsec) image_files : str [N] list of image file names ivar_files : str [N] list of inverse variance file names (default none) edge : int pixel width of apodized edge to use (default 10) kernel, dampsinc, lanczos : parameters for dresample
Returns:	(image, ivar, wcs) : (float32, float32, WCS) [nx, ny] image [nx, ny] image of estimated inverse variance WCS object

Notes

Uses a weighted combination with apodized edges of images to

ensure smooth transitions.

Uses inverse variance to estimate final inverse variance, but

doesn’t track covariance and doesn’t use inverse variance as a weight.

Calls dwcsproject for projection, which calls dresample. Assumes WCS for images and inverse variance files are identical;

otherwise unpredictable results!

dimage.combine.resample(image, x, y, kernel='dampsinc', dampsinc=2.47, lanczos=2.0)[source]¶

Resamples a two-dimensional ndarray image in x, y

Parameters:	image : np.float32 [nx, ny] 2-D ndarray x, y : [N] locations for samples kernel : Kernel to use in resampling (default ‘dampsinc’) dampsinc : Gaussian scale used for ‘dampsinc’ (default 2.47) lanczos : Lanczos scale parameter “a” (default 2)
Returns:	samples : [N] samples

Notes

Calls dresample.c in libdimage.so

dimage.combine.wcsgrid(wcs, nx, ny)[source]¶

Create grid of RA/Dec corresponding to WCS pixels

Parameters:	wcs : WCS object nx, ny : size of image referred to in each dimension
Returns:	(ra, dec) : each member of tuple is [nx,ny] grid of RA/Decs for each pixel

dimage.combine.wcsproject(image, weight, wcs, ra, dec, kernel='lanczos', dampsinc=2.47, lanczos=2.0)[source]¶

Resamples a two-dimensional ndarray image in RA, Dec

Parameters:	image : [nx, ny] 2-D ndarray weight : [nx, ny] weight 2-D ndarray wcs : WCS object ra, dec : [N] locations for samples kernel : Kernel to use in resampling (default ‘dampsinc’) dampsinc : Gaussian scale used for ‘dampsinc’ (default 2.47) lanczos : Lanczos scale parameter “a” (default 2)
Returns:	samples, weights, inbox : [M] samples, weights, and indices into ra,dec

Notes

Calls dresample. Will NOT work for very wide angle input images (180. deg or more), because

0/360 case checking is too simple.