A coupled-grid neural network retina for real-time visual processing

A real-time parallel retinal processor is currently being developed. The processor is based on a hybrid analog/digital neural processing element. The processor uses spatiotemporal characteristics of its input images to perform feature enhancement and extraction, motion detection and velocity measure...

Full description

Saved in:
Bibliographic Details
Main Authors: De Yong, M., Eskridge, T., Palmer, A.
Format: Conference Proceeding
Language:English
Subjects:
Computer vision
Concurrent computing
Feature extraction
Motion detection
Neural networks
Neurons
Retina
Shape
Target tracking
Velocity measurement
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:A real-time parallel retinal processor is currently being developed. The processor is based on a hybrid analog/digital neural processing element. The processor uses spatiotemporal characteristics of its input images to perform feature enhancement and extraction, motion detection and velocity measurements, target tracking, and system focal length adjustments. The subsystems which perform the four operations are described.< >
DOI:10.1109/MWSCAS.1992.271158