PDS_VERSION_ID = PDS3 RECORD_TYPE = STREAM DATA_SET_ID = "VG1/VG2-S-ISS-2/3/4/6-PROCESSED-V1.0" OBJECT = TEXT PUBLICATION_DATE = 2006-07-01 NOTE = "Description of the processing steps behind the data files on this volume." END_OBJECT = TEXT END IMAGE PROCESSING OVERVIEW ------------------------- All of the images were processed according to standard procedures using version 5C/16C of the VICAR software package, as distributed by the Multi-mission Image Processing Laboratory (MIPL) at JPL. See http://www-mipl.jpl.nasa.gov/ for more information about the software. Processing was carried out on a Compaq workstation using the OpenVMS operating system; this was used for maximum compatibility with Voyager image processing procedures c. 1990. Every binary data file on this volume (files ending in .IMG or .DAT) is in VICAR format. See DOCUMENT/VICAR.TXT and .HTML for details of this file format. Note that VICAR files have an ASCII header that can be viewed relatively easily. (1) VGRCDCOPY Images were extracted from the source CDROMs (VG_00xx), decompressed and converted to VICAR format using procedure VGRCDCOPY. The input files were named Cxxxxxxx.IMQ and the decompressed files are archived under the name Cxxxxxxx_RAW.IMG. Note that each VICAR header contains details such as the exposure time, filter name, etc. This information is used by subsequent programs in the image processing. (2) Label Corrections A known bug in VGRCDCOPY inserts a scan rate of 1:1 into the header of Saturn images for which the actual scan rate is 3:1. A special routine was written to correct these errors. Based on subsequent processing, it became apparent that a small number of images contained corrupted headers. Those headers were corrected based on information found in the index files (INDEX/IMGINDEX.TAB and .LBL) and from other sources. A few headers could not be corrected, and those images do not appear in this data set in calibrated form. (3) VGRFILLIN Some Voyager images (primarily from Neptune) contain incomplete lines due to limits on the downlink data rate. VGRFILLIN fills in these lines by interpolating between the adjacent lines. This step was included for forward compatibility with the other volumes of this series; it has generally no effect on Saturn images. (4) RESLOC Each Voyager image has a grid of reseau markings, which record the intrinsic (and variable) geometric distortion of each Voyager image. The RESLOC routine determines the centroid location of each reseau. It writes this information into two different output files. These files are archived (primarily for documentation purposes) as Cxxxxxxx_RESLOC.DAT and Cxxxxxxx_GEOMA.DAT. The former is nothing more than a tabulation of reseau coordinates; it is used as an input file to VICAR routine RESSAR77. The latter also tabulates the geometrically correct positions for the reseau markings and is used as an input file to VICAR routine GEOMA. (5) RESSAR77 This routine hides the reseau markings in an image by replacing them with the average of the nearest pixels. It uses the image and the *RESLOC.DAT file as inputs. This routine also removes a set of standard blemishes that appear repeatedly at the same location in every image. Those images are archived on volume VG_0038 in the MIPL subdirectory. They are named: BLEMLOC_VG1_NA.DAT: Voyager 1, narrow-angle camera. BLEMLOC_VG1_WA.DAT: Voyager 1, wide-angle camera. BLEMLOC_VG2_NA.DAT: Voyager 2, narrow-angle camera. BLEMLOC_VG2_WA.DAT: Voyager 2, wide-angle camera. (6) ADESPIKE This routine erases spikes from the data. It does so by identifying pixels that differ by too much from their neighbors; such pixels are replaced by the average of the nearest neighbors. At this stage in the processing, the resulting images are archived as Cxxxxxxx_CLEANED.IMG. Such images are much simpler to analyze than the corresponding RAW images, although most changes are cosmetic. A comparison between the two can be used to identify unambiguously what images have been changed. (7) FICOR77 This step converts the raw pixel values (DNs or "data numbers") to quantities that are directly proportional to reflectivity, I/F. It does so as follows. First, it subtracts a "dark current" image, which was obtained by reading out the vidicon using the same readout mode as the image in question, but without having opened the shutter (or, in some cases, by having pointed the camera at dark sky). See the appendix below for a discussion of the dark current files used in this data set. Each dark current was selected to match the camera and readout rate of the image being calibrated (in addition to other considerations as noted in the Appendix). When multiple, suitable dark currents were available, the dark current taken closest in time was used. FICOR77 then converts the values to a quantity proportional to I/F. In this step the single-byte, unsigned pixels are converted to two-byte signed integers, which have an intrinsic range from -32768 to 32767. Most commonly, the nominal proportionality constant is such that a calibrated DN = 10,000 corresponds to I/F = 1. However, for some longer exposures of fainter targets, DN = 10,000 corresponds to I/F = 0.1 or even 0.01. These scalings were chosen to ensure that none of a raw image's dynamic range was lost in the calibration step. See DOCUMENT/TUTORIAL.TXT for further details about the absolute calibration of Voyager images. This conversion employs a calibration file that records the responsivity of each individual pixel in the Voyager vidicon. The calibration files are archived on this volume in VICAR format in the MIPL subdirectory on volume VGISS_0038. The files are named: FICOR77_VGn_cc_fff.DAT where n = 1 for Voyager 1 or 2 for Voyager 2; cc = NA for the narrow-angle camera or WA for the wide-angle camera; fff = the filter name, one of CLEAR, ORANGE, GREEN, BLUE, VIOLET, UV, CH4_U, CH4_JS, or NAD (sodium). Note that calibration files for two of the more obscure filters, CH4_U and NAD, are not available for the Voyager 1 wide angle camera. For these images, we used the corresponding Voyager 2 calibration file instead. However, because the absolute gain was different for the cameras on Voyagers 1 and 2, these images almost certainly are not accurately calibrated in absolute terms, although their pixel-to-pixel relative calibration should be reasonable. FICOR77 employs a final step that is intended to refine the absolute calibration. This involves a straightforward multiplication by a scale factor found in input file VGRSCF.DAT, also found in the MIPL subdirectory on volume VGISS_0038. However, see TUTORIAL.TXT for further information about calibration accuracy. The calibrated images are archived on this volume under the names Cxxxxxxx_CALIB.IMG. (8) GEOMA The final processing step is a geometric correction, in which the calibrated image is reprojected onto a new 1000x1000 pixel grid such that the reseau markings fall at their proper locations. This program takes the RESLOC output file Cxxxxxxx_GEOMA.DAT as one of its input file, in addition to the calibrated image file Cxxxxxxx_CALIB.IMG. (9) File translations As discussed above, four image files are associated with each individual Voyager image. Those files are: Cxxxxxxx_RAW.IMG raw image. Cxxxxxxx_CLEANED.IMG reseau markings and spikes removed. Cxxxxxxx_CALIB.IMG calibrated. Cxxxxxxx_GEOMED.IMG calibrated and geometrically corrected. In addition, we have archived the two critical intermediate data files generated by VICAR routine RESLOC: Cxxxxxxx_RESLOC.DAT reseau location table. Cxxxxxxx_GEOMA.DAT geometric reprojection model. These files are provided just in case someone with access to VICAR chooses to reprocess one or more Voyager images. Nevertheless, these files contain useful information about the geometry of the images. For that reason, these files have been converted to easy-to-read ASCII tables. They can be found in this archive under the names Cxxxxxxx_RESLOC.TAB and Cxxxxxxx_GEOMA.TAB. APPENDIX: DARK CURRENT FILES ---------------------------- A critical component of the calibration of Voyager images is to identify a reliable dark current frame to use for each image. The effective dark current underneath each Voyager image varies depending on a number of parameters: Camera: narrow-angle or wide-angle. Spacecraft: Voyager 1 or Voyager 2. Scan rate: The readout speed of the vidicon, which can be 1:1, 2:1, 3:1, 5:1 or 10:1. Shutter mode: In BOTSIM and BSIMAN modes, the cameras are shuttered simultaneously but the narrow-angle camera is read out first. During this delay, the wide-angle can continue to build up dark current signal. Time: The dark current is time-variable, depending on instrument temperature and other factors. For this reason, it is always preferable to use a dark current taken as close as possible in time to the image being calibrated. Because a typical dark current frame contains noise, the subtraction step increases the noise in the image being calibrated. Particularly for images with low signal-to-noise, this is undesirable. One solution is to coadd multiple dark current images prior to the subtraction. For this data set, we have created what we regard as a definitive collection of dark current frames. The Saturn dark currents are found on volume VG_0038 in directory SUMDARKS. We have generated these images by collecting every dark current and sky frame obtained around each flyby, as well as a few wide-angle images of distant targets (for which the target only occupies a few pixels). These images were then grouped by camera, spacecraft, and scan rate. Wide-angle images taken in BOTSIM and BSIMAN mode were handled separately because of their unique properties. In addition, some Voyager images have been edited, meaning that less than 800 samples from each line were transmitted to Earth. Edited dark currents are suitable for the calibration of other edited images, and we occasionally needed to use them for that purpose. We processed each candidate dark current image to the "CLEANED" level and then examined it visually. In most cases, images with corrupted pixels, visible stars or background objects were rejected from further consideration. The remaining images were grouped into sets taken relatively close in time and coadded. See the file INDEX/SUMDARKS.TAB for a table of the files that were coadded to generate each dark current file. This same information can be found in the PDS label of each file of the VGISS_0038 SUMDARKS directory. Before coadding the images, each image was converted from one-byte to two-byte integers and then scaled by a factor of 100. This reduced the loss in precision that arises when integer values are averaged and then rounded back to the nearest integer. During the FICOR77 step discussed above, the dark current images were scaled back down by this factor of 100. In some cases, the number of available dark current images in a particular mode was quite small. In these cases, we had to include imperfect images in the coadding. Corrupted or otherwise unusable pixels were flagged first, and then excluded from the subsequent averaging. In this way, we were able to generate a quite reasonable dark current for every existing image.