|
Right. I want to take a bunch of images, and for each image specify some X/Y coordinates for a rectangle (the same coordinates for all the images), match this rectangle to another image (the same image for all the input images), then if and only if it matches, fill it with white. I cannot for the loving life of me find any remotely sane way of doing this. The best potential solution I've found is maybe using ImageMagick, which seems to be leading me down a nightmarish rabbithole filled with Cygwin and Bash scripts (windows user here). I would settle for a solution which just checked if my specified rectangle had a certain percentage of black pixels, rather than matching it to a second image, honestly. Please dear God someone help me, I've been beating my face against this all loving day. The_White_Crane fucked around with this message at 19:22 on Mar 24, 2017 |
#
?
Mar 24, 2017 19:16
|
|
|
|
| # ? Apr 25, 2024 12:20 |
|
|
ImageMagick has a native windows version (MinGW32) so you don't have to fart about with bash or cygwin, though if you're on Win10, it now has a native linux subsystem which works a lot nicer than cygwin. I'd probably go about this by having imagemagick crop to the square and output the image as uncompressed .bmp to stdout and then pipe that into md5sum which you can then easily compare to your reference image which has been processed in the same way. This method will only work if they are pixel-perfect accurate, same colour depth and same resolution though.
|
|
#
?
Mar 25, 2017 00:10
|
|
|
Yeah, I tried the native windows version and it didn't want to play ball. Then all the advice I got from their community forum was "Now run this Bash script..." In the end I solved the problem using photoshop actions and some javascript so horrible it probably merits ritual suicide.
|
|
#
?
Mar 25, 2017 17:29
|
|
|
If you need something a tad more robust: http://docs.opencv.org/3.0-beta/doc/py_tutorials/py_imgproc/py_template_matching/py_template_matching.html
|
|
#
?
Mar 30, 2017 21:53
|
|
