Analyzing captured image

I need to search in an image for the number and position of pixels of a certain color.
As the screen changes quickly and I need to make sure that I get a consistent state, I thought of first capturing the image and then analyze it for the pixels of the certain color.

The color matching part - with some pre-captured test image - was easy:

set width to imageSize(testImage).x
// loop over whole width of the (first line) of testImage, to find points of the relevant TestColor.
	repeat width-1 times
		set point to (repeatIndex(),1) 
		set currentPointColor to imageColorAtLocation(testImage,point)
		if currentPointColor equals TestColor then
			insert point.x into PointList
		end if
	end repeat

This works great, if testImage is an image I have already in the suite.

But how do I get an up-to-date image to analyze? If I capture it with CapureScreen, I’m not able to work with it afterwards, as something like

set testImage to file (ImagePath & "xyz.tiff") // load image xyz... 

fails with STFileException File /xzy.tiff couldn’t be read using the defaultStringEncoding (UTF8)

How can I access such a dynamically created image?

Or should I use some standard search functionality of Eggplant instead? It is important that I can provide the pixel color as value, not as an pre-captured image…


I would suggest you use the colorAtLocation() function to read the colors directly from a region of the screen. So your code might look something like this:

-- locate the topLeft and bottomRight corners of area to search
set topLeft to imageLocation("topLeft")
set bottomRight to imageLocation("bottomRight")

-- check each pixel in the search area for matching colors
repeat with y=topLeft.y to bottomRight.y
  repeat with x = topLeft.x to bottomRight.x
    if colorAtLocation(x,y) equals TestColor then
      insert (x,y) into PointList
    end if
  end repeat
end repeat

Will that do what you want?


Yes, that’s basically what I want to do.
As the image on the SuT screen rapidly changes, I fear that I get already some changed content for analyzing the last pixels compared to the content I used to analyze first place (it’s approx. 600-800 pixels I need to compare). This would lead to an inconsistent result. That’s why I thought capturing the image and analyze afterwards would be the better solution – but can’t get that to work.


For the captured image approach don’t try to accessing it as a file container. Just use the path.

capturescreen (name:"/tmp/MyScreenShot", rect:(0,0,100,100))
put imageColorAtLocation("/tmp/MyScreenShot.tiff", (50,50))

That was too easy… it works fine… thank you!
Andre :smiley: