PDS_VERSION_ID = PDS3 RECORD_TYPE = STREAM OBJECT = TEXT PUBLICATION_DATE = 2018-08-03 INTERCHANGE_FORMAT = ASCII NOTE = "Description of xpsf directory contents." END_OBJECT = TEXT END xPSF Directory: --------------- This directory contains VICAR-format images corresponding to the filter-specific point spread functions for the ISS Narrow and Wide- Angle Cameras. File naming convention: xpsf_[cam]_[filt1]_[filt2].IMG: Contains the final xPSF (1024x1024) @ 1px resolution in VICAR format xpsf_[cam]_[filt1]_[filt2]_core.IMG: Contains the core of the PSF (4000x4000 for NAC, 2000x2000 for WAC) @ 0.1px resolution in VICAR format The following notes are adapted from the README.TXT file describing the extended PSF analysis and resulting data files produced by Daiana DiNino, 2009. Original PSF analysis by Emma Birath, 2006. Extended wings reduction by Bob West, 2018. Additional cleanup and WAC core-to-xPSF conversion for select filters by Ben Knowles, 2018. ----------------------------------------------------- ----------------------------------------------------- EXTENDED-PSF PROJECT ----------------------------------------------------- ----------------------------------------------------- GOAL: Use satellite images to extend point spread function out to 512 pixels from center for all Cassini ISS filters. ANALYSIS: ---------- We had PSF models that extended to only 200 pixels from the center of the PSF for the NAC filters and 100 pixels for the WAC filters (from the original PSF analysis by Emma Birath, 2006). In order to extend the PSF model to larger radii, WITHOUT modifying the core of the PSF, we: 1) analytically extrapolated the existing PSF models; 2) selected icy satellite images at phase-angle less than 30 degrees (for NAC filters) and less than 45 degrees (for WAC filters) and diameters from a few tens of pixels to a couple of hundred pixels (LOSSLESS compression only); 3) performed standard image calibration (using CISSCAL v3.6beta); 4) ran an IDL code that: - finds centroid (to take care of the fact that in general the satellite image will not be at the center of the image) - calculates S/N and fits S/N as a function of distance (from Moon center). The resulting S/N is used to weight the contribution of each image to the improvement of the PSF. Pixels where the S/N is less than 1.9 are assigned a weight of 0. - removes background noise (using MR software) - iteratively (max = 3) deconvolves each denoised-image with previously known PSF (using MR software), then applies a cut-off to each obtained deconvolved-image and convolves back each deconvolved-image with the known PSF. Finally modifies the PSF by comparing each obtained convolved image with the original denoised-image and weighting this ratio according to the S/N. 5) ran an IDL code that: - sums the deconvolved images (obtained in general with the third iteration of the previous code) and uses the radial profile of this summed image compared to an ideal radial profile to further correct the PSF. One or two iterations are usually enough. 6) created final VICAR files (two for each filters): one containing the whole extended PSF at the resolution of 1 pixels and one containing the core of the PSF at the resolution of 0.1 pixels. All files are currently stored in xpsf/ NOTES: ------ * The final PSF vicar files are oriented the same way as the source VICAR images * No suitable satellite images were found for the following filter combinations (for which we have the core of the PSF though): WAC_CB2_CL2, WAC_CB3_CL2, WAC_CL1_HAL, WAC_IR2_CL2, WAC_IR2_IR1, WAC_IR4_CL2, WAC_IR5_CL2, WAC_MT2_CL2, WAC_MT3_CL2 For these filters, extended PSFs were created by finding the average ratio of the core to the overlapping region of the xpsf for filters where both were available, applying this to the core data, and then filling in the extended region with an average of the existing WAC xpsfs, scaled to the above at the borders. * No core or extended PSFs yet available for: NAC_CL1_CB1, NAC_CL1_MT3, NAC_UV2_CL2 * Subsequent analysis of Titan images by Bob West suggested extended wings too high for all filters. A correction has been applied to account for this.