This application is another example of what can be done with machine vision. It uses the tools from FSI Automation XCaliper. First it locates all objects and then moves the search tool to each object to analyse it. Also notice that the objects can be anywhere in a large area and can be at any orientation, as shown below !

If the program can't find any match to a known animal, the 'unkown animal' message is displayed. Also the message or another signal can be sent out via RS-232 hard wire, or whatever is necessary.

Animal Crackers: a case study in Pattern Recognition This is a demo of Blob tools and Pattern Recognition. Pattern recognition is used in many applications: sorting items, identifying items, finding specific locations on parts for example the registration marks on printed materials or the fiducial points on a printed circuit board. Pattern Matching is used in making Optical Character Recognition applications (OCR). Pattern Matching can be programmed using a variety of techniques including neural nets and gray scale correlation. Most machine vision software packages have a pattern recognition function or add-on module. The user, systems integrator of machine vision systems teaches the System what pattern to look for and what to call the pattern.
Typically, the system has to be shown examples of the pattern and examples of what the pattern is not. In this example, we've taught the System what each different animal cracker looks like: elephant, donkey, cat, etc. In some systems, the system will automatically identify the pattern regardless of rotation. In others the user has to teach the system the various angles, or the system has to teach itself what the pattern looks like at the various angles, both of which require the System to search for multiple patterns which takes processing time during use, or the System has to rotate the image or pattern during run-time which also slows down processing. To speed up processing this application uses a blob tool to first locate items of sufficient size. Then it searches only at the location of the blob and rotates just there. The system outputs the identification, location, and rotation of each object. Objects which are unidentifiable are defects. Applications could include sorting, pick and place, and identification to ensure that the correct part is being used in an assembly.