Note: Abandoned 0.49-0.51 - switching to EDR archives was a too much to bite off and it slowed momentum on the project. So marked that as the 'isis' branch and backed up to 0.485 and continued on with that in the 'main' branch. Will revisit the isis branch someday, when get more into mapping images to spheres.
- repaired
vg download
,vg unzip
,vg convert
, andvg adjust
steps - started Dockerfile to build a Docker image containing step03_convert images, kernels, Linux programs, etc
- Add
vg denoise
step - black out bottom and right 3 pixels, fill in single pixel horizontal lines, black out rectangular blocks - Add
db/denoising.csv
file to control turning denoising step off for certain images (e.g. moons orbiting Uranus, faint rings) - Add
brightness.csv
file forvg adjust
step - override histogram stretching for certain files where noise throws off the brightness adjustment. (first try ignoring 255 values) vg center
in progress
vg download
,vg unzip
handle EDR archivesvg import
converts EDR IMQ files to ISIS CUB files in reorganized folders, with newer PDS volume organization (e.g. volume 5101)
- Switching to Linux and EDR archives so can use ISIS -
vg download
downloads EDR volumes 1-33 now
- Working on
vg map
- preliminary results working okay - could project Jupiter approach frames into cylindrical map, though without photometric corrections. Decided to switch to using ISIS.
- Add
vg crop
command andcrops.csv
to crop/zoom in on images - Add
vg pages
command to generate intro, epilogue, credits, etc.
Added crop/zoom images into Voyager 1 Jupiter encounters - Io volcanoes, best moon images, etc.
vg movies
can assemble movies from arbitrary segments for more editorial control, e.g. mix wide angle with narrow angle segments
Made Voyager 1 Jupiter encounter movies
- Add
-align
option tovg composite
- attempts to align channels using feature detection and matching - Fix
vg adjust
brightness enhancement to ignore hot pixels unless small moon. Improved brightness of dark moon pics and eliminated posterized look from some images due to 16-bit to 8-bit conversion. - Add
vg clear <step> <vols>
command to remove folders, as -y option sometimes fails due to Windows having a lock on a file, e.g. thumbs.db.
Made Voyager 1 Io encounter movie - cleaned up from prior version
- Add
vg inpaint
step to fill in missing or whited out parts of target, where possible - Frame rate constants can be set by System-Craft-Target-Camera
- Add
vg test composite
to test compositing routines, with several examples for validation
Made Voyager 1 Jupiter system movie - lots of edits to composites.csv and framerates.csv
Jupiter rotation segment still a bit unstable - cloud and moon composites not aligned, except where centered.
- Add optional
vg annotate
step - annotate images with imageId, date/time, distance (km), NOTE field text vg clips
can include additional images after frames, as specified inadditions.csv
- use to add hand-tuned mosaics, etcvg adjust
ignores brightest n pixels before doing histogram stretch, to avoid hotspots keeping image dim- Remove
vg clip
bw/color options - all clips will draw from composite step, which will include single channel 'composites' - keeps pipeline simple
vg center
:- return to previous role - will append new center information to
centers.csv
for now - don't try to center image if target size is larger than some threshold (replaces existing
centering.csv
file) - don't center image if includes 'search' in NOTE field - avoids centering ring/satellite searches
- use new
centering.csv
file to turn centering on/off for specific images - overrides above settings - use
framerates.csv
to change frame rates per image - can use sticky ID to set it for a target until it's changed
- return to previous role - will append new center information to
vg clips
framerate depends on angular size of target and target-specific constant, set intargets.csv
vg target
can take a targetpath or volume range
Jupiter rotation movie is still a bit unstable - needs another pass. But Uranus looks fairly good.
Made Io approach clip, Uranus system movie.
vg init centers
:- uses expected target size to narrow down Hough circle search for more accurate results - uses SPICE positions
- uses adaptive thresholding before running blob detection - works better than plain thresholding for pointlike targets
- reduces Hough Canny edge detector threshold if can't detect a circle in case target is dim - helps a lot
- aligns image to expected target disc - works better than aligning to any prior image
- if can't find a circle, fall back to the blob bounding box - stabilization can often handle the rest, so partial targets can be stabilized, even limbs
- optionally draws expected target size on images, based on positions.csv
vg test
draws expected target size on centered images, based on positions.csv- Commands can write output to logfile 'log.txt'
Filled in lots of mis-centered images, but more small stabilization problems than v0.41 - need to align final few pixels
Made another stabilized Voyager 1 Jupiter rotation movie (color)
- Specify composite images with color weights and x,y offsets
- Add
vg init centers <vol>
command - writes stabilized centers tocenters.csv
vg center <vol>
andvg center <imageId>
now just usecenters.csv
andcentersOverride.csv
- Add
vg grab
command - Changed stabilization routine so it stabilizes against every 10 good frames, instead of against previous frame, as Jupiter tended to drift to the left. Had also tried stabilizing it to first image in sequence, but jittered towards the end. Didn't retest it on Uranus.
Still fairly jittery - noise/gaps cause stabilization to stutter
Made stabilized Voyager 1 Jupiter rotation movie (color)
- Add ECC (Enhanced Correlation Coefficient) stabilization [13] to
vg center
step to align centered images more accurately
Made Uranus system movie (color/bw)
- Update
vg center
to use records in centers.csv, if available - Option to use jpeg intermediate files to save space and speed development
- Add
vg test
command to test center detection - Change commands to verbs
- Add
vg retarget
command to print out new retargeting records - Add
vg adjustments
command to separate adjusting and centering images into separate steps
Made Uranus movies (color)
- Combine clips into single movies (eg for Neptune), then a movie combining all movies
Made Uranus system movie (bw)
- Control movie speed with
db/framerates.csv
file
- Handle wildcards and ranges in commands, eg
vg images 5101-5120
,vg images 51*
- Add
vg list
command to show status of volumes - Add -y option to overwrite existing data for a step
- Retarget rings to the main planet so they're included with the appropriate movie
Made Uranus bw and color flyby movies
- Handle relabelling of multitarget images, eg a file may be labelled Titan but it gets centered on Neptune
- Add titles for each movie segment
Made Triton flyby movie bw
- Improved Triton approach centering - blob detection was focusing on pixel-wide edge discrepancy.
- Handle movie targets like
Neptune/Voyager2/Triton
, or just//Triton
- Passing 25/31 (80%) of edge case tests
- Better small/point-like detection with blob detector below 12x12 pixels, before Hough circle detector used
- Use db/centers.csv file to turn off centering at closest approach and slow down movie (currently only Neptune data available)
- Faster movie creation
Made slightly better movies for Neptune flyby, both b&w and color, incl Triton.
- Added command line interface
- Added target discrimination - sorts images and movies into folders based on planet, spacecraft, image target, and camera
- Uses Hough circle detection for centering - still fairly jittery, esp for small circles and crescents
- Uses CALIB images, which have more contrast and darker backgrounds, which helps with circle detection in Neptune images
- Preliminary handling of automatic colorization of frames and movies
Made rough movies for Neptune flyby from volumes 8201-8210, both b&w and color
- No command line interface
- Able to piece together a movie from complete volumes, but no target discrimination
- Uses Blob detection and Hough circle detection for centering
- Uses RAW images, which worked alright for some of the Jupiter images, but not Neptune, which has brighter backgrounds
Made b&w movie for Jupiter approach from volumes 5104-5105