Scalable Data-Parallel Implementations of Object Recognition Using Geometric Hashing

A Robust Neural Network Based Object Recognition System and Its SIMD Implementation

“…However, the use of parallel architectures gives rise to new problems concerning the choice of the modality of parallel computation (SIMD, MIMD) and the most efficient programming language. Current literature on parallel object recongition essentially deal with graph matching approaches implemented in parallel [1] on tree search algorithms [5], geometric hashing [2,16] and parallel hypothesis generation [15]. Here we show that a complete set of computer vision tasks implementing filtering, feature extraction and statistic classification can be efficientely implemented as a whole system on a massively parallel SIMD architecture.…”

Random Data Accesses on a Coarse-Grained Parallel Machine I. One-to-One Mappings

“…A two-phase strategy for performing dynamic permutations on hypercubes using randomization was presented in [25]. A two-phase algorithm that performs fixed (or static) permutations of size n on p processors in O(n/ p) time was presented in [17,18].…”

Portable and Scalable Algorithm for Irregular All-to-All Communication

“…The exact communication times are shown based on the experimental results (see Section 4). The flat model is widely used in analyzing communication time on state-of-the-art HPC platforms [3,15,21,22,30]. Note that, when a collection of messages is transferred between a pair of nodes, the throughput can be increased by pipelining the communication steps.…”