Abstract: A computerized method and system for non-intrusive testing of an information processing system. This includes stimulating a system-under-test, capturing an image from the system-under-test visual output and one or more of the following: converting the captured image to a grey-scale bitmap, deriving pixel patterns from sub-portions of the captured image, normalizing the derived pixel patterns, scoring pixels in a pixel pattern according to color intensity, comparing pixel patterns with graphical object definitions, and finding and outputting matches taking into consideration tolerances for color, resolution, object spacing and overlap, and font kerning variation. The output is optionally a text string representing the recognized text within the captured image. The method and system also automatically learns new graphical object definitions when appropriately provided with graphical objects.