Share Email Print

Proceedings Paper

Fast parallel algorithms: from images to level sets and labels
Author(s): H. T. Nguyen; Ken K. Jung; Raghu Raghavan
Format Member Price Non-Member Price
PDF $14.40 $18.00

Paper Abstract

Decomposition into level sets refers to assigning a code with respect to intensity or elevation while labeling refers to assigning a code with respect to disconnected regions. We present a sequence of parallel algorithms for these two processes. The process of labeling includes re-assign labels into a natural sequence and compare different labeling algorithm. We discuss the difference between edge-based and region-based labeling. The speed improvements in this labeling scheme come from the collective efficiency of different techniques. We have implemented these algorithms on an in-house built Geometric Single Instruction Multiple Data (GSIMD) parallel machine with global buses and a Multiple Instruction Multiple Data (MIMD) controller. This allows real time image interpretation on live data at a rate that is much higher than video rate. The performance figures will be shown.

Paper Details

Date Published: 1 July 1990
PDF: 15 pages
Proc. SPIE 1246, Parallel Architectures for Image Processing, (1 July 1990); doi: 10.1117/12.19577
Show Author Affiliations
H. T. Nguyen, Lockheed Palo Alto Research Lab. (United States)
Ken K. Jung, Lockheed Palo Alto Research Lab. (United States)
Raghu Raghavan, Lockheed Palo Alto Research Lab. (United States)

Published in SPIE Proceedings Vol. 1246:
Parallel Architectures for Image Processing
Joydeep Ghosh; Colin G. Harrison, Editor(s)

© SPIE. Terms of Use
Back to Top