I am automating remote machines through a vnc client. To detect menu screens I initally tried using tesserat to read the text but this proved very slow and outputs were too variable. I moved to using pixel checksum on specific areas of the screen to determine the scripts location in the menu. This works most of the time..
My issue is that even with a very good VNC connection there are occasional small artifacts which cause Pixelchecksum to produce different values. I was trying to find a more imprecise way to compare the regions. Ideally a comparison where I could provide and accuracy value such as 98% allowing upto 2% of the region to contain artifact.
I could create a pixel array of the region and compare to a perfect sample and produce an output of the percentage match. I think coding my own version is likely to add significant delay into the script so was hopeing there was some library or function I've missed that already does this or something similar.
Thanks Kyu
EDIT: I found this http://www.autoitscript.com/forum/topic/56423-compare-pictures/page__hl__compare basically the implementation of my idea, still would like a faster way
My issue is that even with a very good VNC connection there are occasional small artifacts which cause Pixelchecksum to produce different values. I was trying to find a more imprecise way to compare the regions. Ideally a comparison where I could provide and accuracy value such as 98% allowing upto 2% of the region to contain artifact.
I could create a pixel array of the region and compare to a perfect sample and produce an output of the percentage match. I think coding my own version is likely to add significant delay into the script so was hopeing there was some library or function I've missed that already does this or something similar.
Thanks Kyu
EDIT: I found this http://www.autoitscript.com/forum/topic/56423-compare-pictures/page__hl__compare basically the implementation of my idea, still would like a faster way
![:)](http://aut1.autoit-cdn.com/forum/public/style_emoticons/default/smile.png)