Image Doctor

Has anyone had success using Image Doctor. It does not seem to be working to well for me. For instance I have a script that I run on windows and now I would like to test it on Linux. Our produt looks pretty similar on both SUT’s. When it can’t find an image it comes up with alternatives that aren’t even close. A standard button for selecting a File tab (Captured image The word “File” with Gray background on windows) on the Linux host it can’t find it but when image doctor runs it comes up with the Letters “Me” from the word memory on the completely opposite end of the screen for the Discrepancy search and 2 locations under the Dynamic Tolerance that are just gray portions of the screen. Is there a more detailed description (other thatn the manual as to what is going on. how would it think ME was a valid choice for the word “File” Any Ideas.

Thanks,
Ray

Please keep in mind that the Image Doctor does an excellent job of matching what it thinks are image elements close to that which it is told to compare. You can find some information also in the Release Notes as well. “me” and “file” could be seen as similar disregarding the text aspects of the image. In this case the three vertical elements of the letter “m” could be easily matched in the tolerance or discrepancy aspect settings of the original image.

The Image Doctor works by trying a series of increasingly agressive algorithms to try to match your image.
[list]
[]Dynamic Tolerance will gradual increase the color difference that is allowed in the image matching.
[
]Alternate Types tries the more progressive Tolerance, Pulsing, and Text Image searches
[*]Discrepancy Search allows an increasing number of pixels to completely not match
[/list:u]
Often these approaches work and find the image easily, but none of these are perfect solutions and all are prone to false positives which is why they are part of the interactive doctor (and not just always tried when searching for images).

If the default algorithms don’t find it for you the image doctor still makes it much easier to adjust your test scripts by highlight the correct image in the remote window and choosing Add Representation from the Action Menu to create an Image Collection.