DocumentCode
3410090
Title
Programmable hybrid co-processor for real-time image understanding
Author
Erten, Gamze ; Salam, Fathi M.
Author_Institution
IC Tech. Inc., Okemos, MI, USA
Volume
2
fYear
1995
fDate
Oct. 30 1995-Nov. 1 1995
Firstpage
885
Abstract
A novel hybrid image co-processor architecture that utilizes analog and digital computing techniques cooperatively is described. The chip computes stereo and motion correspondence at a rate of 30 frames/sec. This feature combined with programmable capabilities of a conventional processor makes it possible to execute in real-time many powerful yet so far underutilized computing techniques such as multiresolution or image pyramid processing. The overall system is designed to be compatible with existing computer architectures to allow for commercial product development in the near future. Additional benefits of the system over existing image processing methods are multiple programmable configurations, increased processing speed, low power consumption, reduced size and cost.
Keywords
image resolution; analog VLSI; analog computing techniques; commercial product development; computer architectures; digital computing techniques; edge detection; hybrid image coprocessor architecture; image convolution; image processing methods; image pyramid processing; low power consumption; motion correspondence; multiple programmable configurations; multiresolution; neuronal visual processing; processing speed; programmable hybrid coprocessor; real-time image understanding; reduced cost; reduced size; stereo correspondence; Analog computers; Analog integrated circuits; Cameras; Computer architecture; Computer vision; Coprocessors; Digital integrated circuits; Image processing; Layout; Very large scale integration;
fLanguage
English
Publisher
ieee
Conference_Titel
Signals, Systems and Computers, 1995. 1995 Conference Record of the Twenty-Ninth Asilomar Conference on
Conference_Location
Pacific Grove, CA, USA
ISSN
1058-6393
Print_ISBN
0-8186-7370-2
Type
conf
DOI
10.1109/ACSSC.1995.540827
Filename
540827
Link To Document