Image Search Methods

Image Search Methods search the screen for an object (component) based on template image(s), image collection(s) or object parameters such as the size, colour or shape. The method typically returns a list of locations and/or rectangles on the screen where the specified object was detected. This method group is often employed to automate actions such as "find and click a button" or "make sure that a component is visible on the screen".

The Image Search Methods supported by T-Plan Robot are:

  1. The Image Search V2 (code "search2") method searches the screen for components by template images and collections. The algorithm features three search modes (exact, tolerant and fuzzy) and ordering of results by the lowest difference or location.

  2. The Image Search (code "search") method is an older version of the V2 algorithm. It is also capable of searching the screen for components by template images and collections with a limited tolerance to minor image changes.

  3. The Object Search (code "object") method locates objects by colour, size and an optional sample template image. The algorithm is capable of searching for rotated instances of the object and provides certain robustness against eventual object overlapping.
    Note: The other methods listed above are more suitable for GUI automation, as this method rather targets analysis of graphical systems such as GIS outputs, maps, charts and similar.

  4. The Google Cloud Vision (code "vision") is an experimental method linking to the Google AI. It is supported since Robot 8. It may be used to verify whether certain objects are on the screen.