Age | Commit message (Collapse) | Author |
|
|
|
This reverts commit 7fa913da473caae893b6791d9f5fb10ca1472360.
|
|
|
|
|
|
|
|
|
|
|
|
The implementation should be bug-free now. Of course, I thought
that previously, then found more, so we shall see. There are
more tests, though, and they definitely pass, so that's a good
start.
|
|
general Gray16 image
This is what found the bug which was fixed by commit b9e6c0d3.
|
|
column or row 0 returns 0 for calculations
|
|
and Proportion which was best in the preproc package
|
|
|
|
|
|
general image Bounds functions in Window functions
|
|
|
|
functions do the same job more flexibly
|
|
|
|
|
|
before processing has the same result
|
|
This is intended as an alternative to the WithSq and ToSqIntegralImg
functionality, though the MeanStdDevWindow function will need to be
changed to use it, and it hasn't been heavily tested yet.
|
|
Now the integralimg can be created directly from an image
using image/draw.Draw.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
integralimg package under it
|
|
|
|
|
|
|
|
|
|
This is useful for prebinarised images, which don't need full preprocessing,
but do require wiping, albeit with a more conservative threshold.
|
|
|
|
|
|
for wipe
|
|
|
|
|
|
|
|
ensure connections to new ips not in known_hosts still succeed
|
|
|