|
Title: EXPERIMENTING WITH RECOGNITION ACCELERATOR FOR PAVEMENT DISTRESS IDENTIFICATION
Accession Number: 00728498
Record Type: Component
Record URL: Availability: Find a library where document is available Abstract: A study was conducted using a computer board embedded with an artificial neural network (ANN) microchip for pattern recognition of pavement distress classification. The basic principles behind ANNs and pattern recognition are discussed. The hardware architecture of the Ni1000 recognition accelerator chip, which is the core of the ANN computer board, is presented, and the principle of operation of the restricted coulomb energy algorithm used in the chip is discussed. It is demonstrated that the Ni1000 recognition accelerator chip can be used for pattern recognition of pavement distress. Distresses in pavement images have been successfully classified using the Ni1000 recognition accelerator. The Ni1000 has the potential to be used as the core processing unit for distress classification at highway speeds.
Supplemental Notes: This paper appears in Transportation Research Record No. 1536, Traffic and Pavement Surface Monitoring Issues.
Language: English
Corporate Authors: Transportation Research Board 500 Fifth Street, NW Authors: Nallamothu, SWang, KCPPagination: p. 130-135
Publication Date: 1996
Serial: ISBN: 0309059089
Features: Figures
(7)
; References
(12)
; Tables
(3)
TRT Terms: Uncontrolled Terms: Subject Areas: Highways; Pavements; I23: Properties of Road Surfaces
Files: TRIS, TRB
Created Date: Nov 26 1996 12:00AM
More Articles from this Serial Issue:
|