Autosophy Pattern Recognition and Vision

Robot vision systems using parallel omni dimensional networks

Application Proposal: 1998

Abstract: Even in highly predictable environments conventional data processing computers require enormous processing power to achieve real time pattern recognition with effective reliability. Such is a consequence of conventional Shannon-based information theory in which data content is essentially irrelevant. Alternatively a new “Autosophy” information theory defines information by data content or “meaning” such that pattern recognition is not only possible but profoundly inherent in Autosophy-based technologies. Faster and more reliable pattern recognition may thus be achieved within more complex, natural environments. A prototype text learning Autosopher database is already operational and available for demonstration. The system acts like a “black box” which is educated similarly to a human child to learn text data without conventional programming. This project would involve the addition of image learning and real time pattern recognition capabilities. Using Content Addressable Memories search access time to locate a matching image data pattern will be approximately one microsecond per pixel and virtually independent of the database size. This should allow pattern recognition even between real-time data streams and massive reference databases. Applications include the identification of target objects in real time video, part inspection, industrial robot vision, and security checks such as airport surveillance systems.

Anticipated Applications: Autosophy pattern recognition techniques are highly suitable for applications requiring the intelligent storage, indexing, compression, comparison and retrieval of image and video data. These include the instant processing of fingerprints. Search access time to any stored record or image will be virtually instantaneous and independent of the database size. The technology thus has the potential to replace conventional computer databases and evolve into a true robot vision. In Autosopher databases images are stored in a mathematical hyperspace, which exhibits a saturating storage requirement. The more images already stored the less storage space is required to store additional images. For very large databases containing millions of images this could result in many orders of magnitude greater memory storage capacities. Database searching and “mining” is nearly instantaneous and independent of the number of stored images to be searched.

Keywords: Autosophy, Robot Vision, data compression, Databases, Archiving, Memory.

Available downloadable documents:

Proposal document – MS Word doc

Related Publication 1993 – Webpage htm

Related Publication 1992 – Webpage htm