DocumentCode
3318335
Title
Design and implementation of a 3D-LSI character recognition image sensor
Author
Kioi, K. ; Shinozaki, T. ; Toyoyama, S. ; Shirakawa, K. ; Ohtake, K. ; Tsuchimoto, S.
Author_Institution
Sharp Corp., Nara, Japan
fYear
1990
fDate
9-12 Dec. 1990
Firstpage
279
Lastpage
282
Abstract
A four-story structured character recognition image sensor has been implemented with three-dimensional LSI technology, integrating 55-thousand diodes and 0.22-million transistors on a 14.3-mm square single die. Three-dimensional LSIs will achieve very high performance by exploiting the structural parallelism by way of the inherent parallelism of the algorithm. The chip, which is the second-step prototype of the Intelligent Image Sensor, is able to sense 12 characters of an image at the same time and recognize 64 kinds of characters containing the alphabet in upper and lower case, Arabic numerals, and some symbols, each consisting of a 10*14 bit matrix. The present chip allows a 20 times larger degree of data parallelism in image computation and a considerably higher speed of recognition in data matching compared with a previous chip. The efficiency of the chip has been verified.<>
Keywords
MOS integrated circuits; VLSI; character recognition equipment; computerised pattern recognition; image sensors; optical character recognition; parallel architectures; 14.3 mm; 3D-LSI; Intelligent Image Sensor; OCR; character recognition image sensor; data matching; data parallelism; four layer LSI; four story image sensor; image computation; implementation; second-step prototype; single die; speed of recognition; structural parallelism; Character recognition; Concurrent computing; Diodes; Image recognition; Image sensors; Intelligent sensors; Large scale integration; Parallel processing; Prototypes;
fLanguage
English
Publisher
ieee
Conference_Titel
Electron Devices Meeting, 1990. IEDM '90. Technical Digest., International
Conference_Location
San Francisco, CA, USA
ISSN
0163-1918
Type
conf
DOI
10.1109/IEDM.1990.237175
Filename
237175
Link To Document