Computing the Hough transform on an MIMD hypercube
The computational expense of Hough transforms has prompted researchers to investigate the feasibility of parallel implementations. Most such work to date has dealt with special- purpose architectures or with implementations for SIMD machines such as the MPP or GAPP. This report considers the problem of efficiently implementing Hough transforms on an MIMD hypercube architecture. Beginning with a general analysis of how the data structures might be partitioned to allow a parallel computation, we formulate algorithm and data structure partitioning strategies appropriate to the architecture and then discuss modifications to optimize the performance for a particular hypercube system (the NCUBE). We also present the results of implementing and benchmarking Hough transforms on a 64-processor NCUBE. 22 refs., 6 figs., 1 tab.
- Research Organization:
- Oak Ridge National Lab., TN (USA)
- DOE Contract Number:
- AC05-84OR21400
- OSTI ID:
- 5970445
- Report Number(s):
- CONF-8906167-1; ON: DE89013920
- Country of Publication:
- United States
- Language:
- English
Similar Records
Computing the Hough transform on a scan line array processor
Hypercube performance