PC Eyebot inspection tools from Sightech Vision Systems Inc. learn patterns and textures to inspect products for defects.

In addition, if a product's patterns or textures change, the PC Eyebot stores its previous "training" and learns the new patterns for inspections and outputs the differences to PLCs and enunciators for physical responses.

The texture-inspection mode on the PC Eyebot inspection tool from Sightech Vision Systems Inc, San Jose, Calif., lets the device learn and store patterns or texture features. These learned features consist of edges or simple shapes that represent the definitive qualities of an object under inspection.

Such capabilities, reports Sightech, have long been elusive for conventional machine-vision processors.

In one instance, a PC Eyebot detected occasional stray fibers in biomedical filters after first learning the filter-fiber patterns. Then, the PC Eyebot found loose fibers on a few filters as they rotated in front of a 640 480-pixel camera.

Fibers were about the thickness of a human hair and similar in color to filter material. Stray fiber directions differed from the rest of the filter texture and enabled the tool to consistently identify those unsatisfactory filters.