This file contains information helpful for utilizing improved DISR images. The originally archived DISR images suffered from at least two deficiencies. 1) Movement of the fiber-optic transmittance pattern as projected onto the detector caused the on-board flat fields to incorrectly compensate, and 2) Rounding of the compression algorithm's Discrete Cosine Transfer coefficients upon reconstruction introduced unnecessary additional compression artifacts. Erich Karkoschka has developed an improved method of processing the DISR images that compensates for some of the deficiencies. Improved images are available in the archive under: \EXTRAS\PROCESSED_IMAGES, with an explanation of the processing steps. We also now include elements of these improved images to allow the user more flexibility in utilizing the DISR images. These elements are archived in the EXTRAS directory as: \EXTRAS\IMAGE_ELEMENTS. The following steps are executed when creating a DISR image... 1) The 12 bit image set (HRI, SLI & MRI) is exposed and clocked into memory. 2) Each image is divided by the appropriate flat-field map (HRI, SLI & MRI). 3) Known bad pixels are replaced by adjacent pixels (via the bad pixel map). 4) The images are reduced from 12 bits to 8 bits by square root algorithm. 5) The images are hardware compressed 6) The images are transmitted to Earth 7) The images are decompressed 8) The images are converted back to 12 bits using de-square root algorithm. The resulting images are contained in the DISR archive (\DATA\IMAGE). Obviously steps 1 through 5 are occur on the spacecraft, however there is some flexibility on how the images are processed after transmission. Decompressing the images... Received on Earth are the Discrete Cosine Transform (DCT) coefficients. Accurate reconstruction of the 12 bit images is vital, since uncompressed 12 bit data was used to calibrate the instrument. The first step in using the images is to perform the inverse transform to recover the 8 bit pre-transmitted images. To perform the inverse-transform the quantization number and threshold are provided. File DCT_COEFFICIENTS.TAB contains a table with the image number, the quantization number (i.e. 3 means that the last three binary digits were deleted), and the estimated threshold (i.e. 5 means that quadruplets of amplitudes were deleted if none was above 4.5 in absolute value). The compressor threshold for each image is estimated based on the relative frequency for the three lowest amplitude bins. It is estimated to agree with the actual threshold for more than 90% of the images (none are off by more than 1). Applying the inverse DCT results in the images presented in the \EXTRAS\IMAGE_ELEMENTS\RAW_IMAGES directory. This directory contains 606, 16 bit Portable Grey Map (PGM) formatted images. The data-numbers are the 8 bit values multiplied by 128. The images are rotated 180 degrees from what the imager actually sees (pixel 0,0 is in the upper right corner of the scene). To aid digesting the PGM files, tables of their pixel values are supplied in: \EXTRAS\IMAGE_ELEMENTS\RAW_IMAGES\TABLE_FORMAT Converting to 12 bit depth... The images were converted from their original 12 bit depth to 8 bits using an on-board pseudo-square root transformation via look-up table. To return the images to 12 bits, transformation tables for each image are supplied in the directory \EXTRAS\IMAGE_ELEMENTS\SQRT_TABLES. The tables contain 4 columns: 1) The 8 bit value 2) The lowest 12 bit equivalent 3) The highest 12 bit equivalent, and 4) The number of 12 bit levels corresponding to that 8 bit value. For example the entry for 8 bit level 4 is: 4 7 10 4 which says that 8 bit level 4 maps to 12 bit levels 7 through 10, corresponding to 4 possible 12 bit values (7, 8, 9, & 10). It is readily apparent that this is a lossy transformation. Flat Fielding... Improved flat fields are included for each image in the directory: \EXTRAS\IMAGE_ELEMENTS\IMPROVED_FLAT_FIELDS. There is one flat field (FF) for each image. The FF files are in PGM format (same as the raw images). The FF pixel values are scaled up by 10,000 to avoid using decimal numbers, and thus must be divided by 10,000 before being applied by dividing each image by the flat field. To aid digesting the PGM files, tables of their pixel values are supplied in: \EXTRAS\IMAGE_ELEMENTS\IMPROVED_FLAT_FIELDS\TABLE_FORMAT To use these flat fields the original, on-board flat fields must be removed first. They are also supplied in tabular form in the directory \EXTRAS\IMAGE_ELEMENTS\ON-BOARD_FLAT_FIELDS. The flat fields are removed by multiplying each pixel value in the image by the corresponding value from the table. Note that the on-board flat field contain only 254 rows. The first and last rows of the images were not to be used. If the user desires to attempt to use image rows 0 and 255 it is recommended that the on-board flat field table be extended by duplicating rows 0 and 253. Dark Current... There are at least 4 contributions to non-illuminated charge accumulation (aka dark current): 1) Charge accumulated while the image is being clocked from the 'image zone' to the 'memory zone' of the CCD. 2) Charge accumulated while waiting in the memory zone to be read-out. 3) Charge accumulated in the shift register, and 4) An offset programmed in the data-handling software. The resulting effect is that the image pixel values are offset by a constant amplitude, and a an amplitude that varies along the columns of the images. The column dependent dark current is presented in directory \EXTRAS\IMAGE_ELEMENTS\DARK_CURRENT_IMAGES. These are again PGM formatted images, where the dark current values are multiplied by 8. For images after #243 the detector chip was cool enough that the per-pixel dark current is considered negligible. To aid digesting the PGM files, tables of their pixel values are supplied in: \EXTRAS\IMAGE_ELEMENTS\DARK_CURRENT_IMAGES\TABLE_FORMAT The text file DARK_OFFSET.TXT contains coefficients for the pixel-independent dark current term. The pixel-independent dark data number is 0.9 + 0.25 * value in table. The two columns in the table are 1) the image sequence number & 2) the dark current coefficient. All three images in a triplet have the same coefficient. For additional information about dark current compensation see the description of the image processing (IMAGE_PROCESSING_STEPS.TXT) in \EXTRAS\PROCESSED_IMAGES, and in the DISR Users' Guide. The files RAW_IMAGE_LIST.TXT, DARK_IMAGES_LIST.TXT, and FLAT_FIELDS_LIST.TXT contain summary information about each file in their corresponding directories.