Parallel implementation of associative memories for image classification

• Published in: Parallel computing Vol. 19; no. 6; pp. 667 - 684
• Main Authors: Pagano, Fabrizio, Parodi, Giancarlo, Zunino, Rodolfo
• Format: Journal Article
• Language: English
• Published: Amsterdam Elsevier B.V 1993; Elsevier
• Subjects: Applied sciences; Associative memory; Computer science; control theory; systems; Computer systems and distributed systems. User interface; Exact sciences and technology; experimental results; parallel implementation; real-time associative visual systems; Software; transputer network; Associative memory; experimental results; transputer network; real-time associative visual systems; parallel implementation; Associative storage; Experimental result; Parallel implementation; Real-time associative visual system; Parallelism; Real time system; Transputer network
• ISSN: 0167-8191; 1872-7336
• DOI: 10.1016/0167-8191(93)90014-C

Associative techniques are useful in computer vision because they are botably able to robustify a recognition system. The noise-like coding model of associative memory has been already applied successfully to image-classification. This paper describes the implementation of the associative system on transputer-based architectures. After explaining the model's basic formalism, the paper marks out the key-generation mechanism, the data-mapping strategy, and the hierarchical processor organization. The basic result of this research is a general methodology for efficient HW configuration of real-time associative visual systems. The system's efficiency can be predicted by theoretical derivations, in which both the FFT-computation speed and the data-transmission speed play a crucial role. Experimental results including different HW configuration and different image-sizes always confirmed theoretical expectations.