A High-Speed Object Recognition Chip Based on Biologically-Realistic Hybrid Temporal Processing Element
Abstract
The technical objective of this Phase I effort is to evaluate the feasibility of a binocular object recognition system for the integrated active vision system (AVS) currently under development at IRSI. The AVS based on a custom very large scale integration (VLSI) Hybrid Temporal Processing Element (HTPE) developed by IRSI (patent pending). Active vision implies that the image collection and processing operations form a single, integrated, adaptive system in which high-level processes, such as object recognition, can augment lower- level processes through feedback connections and can itself be controlled by higher levels. For example, the object recognition process can instruct the oscillating saccadic perception system to enhance the image resolution by a particular amount to assist in the object recognition process. Higher level perception systems can guide the object recognition processes by using expectations of what objects might be present in the image. Therefore, the AVS behaves a single image collection and processing system with coupled low-and high-level capabilities
Document Details
- Document Type
- Technical Report
- Publication Date
- May 26, 1993
- Accession Number
- ADA265490