Matt McWilliams
95f93eabc9
Actually draw number
2023-02-11 19:23:39 -05:00
Matt McWilliams
45af6406da
Script for drawing all registration marks
2023-02-11 19:21:38 -05:00
Matt McWilliams
479e9f5207
calibrate.py should actually be template.py. Calibration will be for printers and plotters. Different process.
2023-02-11 19:07:41 -05:00
Matt McWilliams
7581cc75b5
Force commit registration mark
2023-02-10 15:46:50 -05:00
Matt McWilliams
1f572295f6
Trying a contrived version of the affine_poc that demonstrates the issue
2023-02-10 15:36:51 -05:00
Matt McWilliams
672f4212b5
Create masks?
2023-02-09 22:55:42 -05:00
Matt McWilliams
28d959ff1b
Tried more approaches
2023-02-09 22:17:18 -05:00
Matt McWilliams
e560955537
Make script more like proof of concept script. Still nothing.
2023-02-08 15:49:23 -05:00
Matt McWilliams
34e5153948
Moved everything into RGBA from RGB in case alpha channel allowed for proper use. Not working. Cannot understand why proof of concept works but not this.
2023-02-08 15:34:40 -05:00
Matt McWilliams
a1a57ef58c
Image overwriting continues even after loading a valid image and drawing the hole punches on
2023-02-08 10:58:41 -05:00
Matt McWilliams
aabdac74b5
Apply_image.py script is correctly placing images in the template locationg BUT is overwriting the starting blank image with a black background and doing so for all subsequent images. Not sure how to properly address this
2023-02-08 10:42:45 -05:00
Matt McWilliams
89678edd8e
Wrote notes on the spec for the two apply scripts which will put the calibration data into action.
2023-02-07 02:11:46 -05:00
Matt McWilliams
d57a145f1c
Update readme with info about calibration script and documenting its output
2023-02-07 02:11:19 -05:00
Matt McWilliams
80e4d8e251
Create a affineTransform proof of concept script that takes one image, finds points within it and to scale and distort an image to and then places it there.
2023-02-07 02:10:58 -05:00
Matt McWilliams
800b5af94e
Write calibration metadata to a json file that includes the normalization info
2023-02-06 13:02:32 -05:00
Matt McWilliams
1a8491919c
Can now find all 4 frames and correctly orient the rectangles that contain them
2023-02-06 00:13:05 -05:00
Matt McWilliams
2c2604c4b0
Find registration marks, tag center of mark (not top left corner)
2023-01-30 09:33:30 -05:00
Matt McWilliams
1589076974
Work on matching registration marks against document. Segmenting image without copying is causing errors.
2023-01-30 09:26:43 -05:00
Matt McWilliams
adf465fd4c
Select and group registration marks
2023-01-29 22:53:51 -05:00
Matt McWilliams
727713f41b
All work on aktinson dithering-related scripts
2023-01-02 12:21:40 -05:00
Matt McWilliams
dea1fb8134
Calibrate script work, having trouble finding pattern so will have user select
2022-11-24 19:41:01 -05:00
Matt McWilliams
81d79a1602
Write hole punch file as json
2022-11-24 12:10:05 -05:00
Matt McWilliams
ab9e31b49e
Begin the calibrate script which will read the .txt file made by the normalize script. JSON would make more sense.
2022-11-24 11:22:03 -05:00
Matt McWilliams
9b48e52736
Migrate common functions into common.py to share between scripts
2022-11-24 11:21:06 -05:00
Matt McWilliams
91e806f9ab
Add hole punch diagram to readme
2022-11-24 11:19:17 -05:00
Matt McWilliams
b93ef905ae
Remove excessive logging
2022-11-15 20:55:06 -05:00
Matt McWilliams
5a0da25109
Eval files are great, can be used for further stabilization if needed
2022-11-15 14:05:45 -05:00
Matt McWilliams
58955ade4b
Normalize works, I like it. Image size is the real bandwidth limiter but images look good enough. SHould create a script to onionskin all scans for evaluation. Maybe run the holepunch detection on normal and write text file that prints the centers? Will do that now.
2022-11-15 13:59:51 -05:00
Matt McWilliams
a352ad1062
I dropped the stash
2022-11-14 13:46:30 -05:00
Matt McWilliams
89eaa29b81
Why...
2022-11-14 13:45:39 -05:00
Matt McWilliams
163fbad520
okay?
2022-11-14 13:44:23 -05:00
Matt McWilliams
d5a90ca8ae
Fixed?
2022-11-14 13:43:58 -05:00
Matt McWilliams
1a28ccba66
What is happening
2022-11-14 13:43:22 -05:00
Matt McWilliams
6ae5797ebb
Fix sloppy merge
2022-11-14 13:42:44 -05:00
Matt McWilliams
ff04b2aa37
Add a few pattern matching algorithms to notes
2022-11-14 13:40:26 -05:00
Matt McWilliams
2798729373
Work on normalization script
2022-11-13 22:36:25 -05:00
Matt McWilliams
9e11d95447
Credit "borrowed" code used for angle normalization
2022-11-11 10:19:31 -05:00
Matthew McWilliams
66aa5d1712
Add a script that draws the calibration page fiducials
2022-11-11 10:09:14 -05:00
Matthew McWilliams
9ca86eb862
Add an install script for setting up the directories with dependencies
2022-11-11 10:08:49 -05:00
Matthew McWilliams
ca6d7d7d0b
Specified the "analyze" script as the "normalize" script. Can be used for multiple processes.
...
At the point where the straightened image needs to be centered onto an idealized blank sheet.
2022-11-10 18:06:47 -05:00
Matthew McWilliams
602b6dc26c
Draw lines between squared points. Get rotation angles (still need to fix verts_angle())
2022-11-09 16:56:14 -05:00
Matthew McWilliams
e1c6777bf8
Only look at 20% margins for hole punches. Works on all existing examples
2022-11-09 16:55:02 -05:00
Matthew McWilliams
73b61c0cab
Fix appendStr -> append
2022-11-09 09:45:07 -05:00
Matt McWilliams
f103133b9f
Almost blew past a good stopping point.
...
This script can now, on at least one example. locate the largest contours on the page and then put those contours in an array.
It should also look for the largest, most outlying, to prevent errors in the case of large contours in the image.
They should never go past the holes.
Next step: use this for straightening, which will be good for the image extraction script.
Use them as registration points and then just rotate, shift and "shoot" by cropping at a point relative to the holes.
For calibration/templating, look into pattern matching fiducials after getting hole punch coordinates.
2022-11-08 22:47:36 -05:00
Matt McWilliams
e5d9d654ac
Merge remote-tracking branch 'origin/main'
2022-11-08 21:10:50 -05:00
Matthew McWilliams
cd1edbe538
Started analysis script with single example taken from stackoverflow as a starting point. Had to modify it to support opencv 4.x but got it to run. Finds holes.
2022-11-08 19:11:15 -05:00
Matt McWilliams
55cc8ed22c
Add a "stipple_random" script that, unlike the sine, uses shuf and a sequence of integers to generate a random number to apply as the "step" in the range of stipple settings
2022-11-04 21:42:39 -04:00
Matt McWilliams
e8a262e64f
Lift pen before putting in position
2022-10-25 20:12:18 -04:00
Matt McWilliams
fb617cfaee
Progress
2022-10-25 20:07:58 -04:00
Matt McWilliams
e1451abbe5
Use --no-adjust to prevent page from resizing
2022-10-12 13:49:08 -04:00