What do you think?
Rate this book
On a Parallel Implementation of Geometric Hashing on the Connection Machine: Technical Report 554
Excerpt from On a Parallel Implementation of Geometric Hashing on the Connection Machine: Technical Report 554
In geometric hashing, the collection of models are used in a preprocessing phase (executed off-line and only once) in order to build a hash table data structure. The data structure encodes the information about the models in a highly redundant multiple-viewpoint way. During the recognition phase, when presented with a scene and extracted features, the hash table data structure is used to index geometric properties of the scene features to candidate matching models. A search is still required over features in the scene. However, the geometric hashing scheme no longer requires a search over the features in the model sets. The result is that the recognition phase offers computational efficiencies over more traditional model-based vision methods. As we describe elsewhere there is parallelism available in the search over scene features, and there is parallelism in the indexing process inherent within the hashing scheme. Further, the computational efficiencies afforded by geometric hashing translate into a reduction in the number of independent tasks that may be simultaneously conducted (at the expense of increased storage requirements), thereby decreasing the number of processors that are needed in a parallel implementation.
In this paper, we report on an implementation on the Connection Machine of one of the algorithms described in We also describe a number of modifications that are useful for efficient parallel implementation of the method for the particular case of point features. The last-section of the paper presents the results of a number of experiments that were carried out using this implementation. These experiments had a two-fold purpose: first, we evaluated the real time performance of the implementation with a large database of models; second, the behavior of the method in the case of rigid or similarity transforms was examined, as a function of the noise present in the input. Other statistics are also presented.
About the Publisher
Forgotten Books publishes hundreds of thousands of rare and classic books. Find more at www.forgottenbooks.com
This book is a reproduction of an important historical work. Forgotten Books uses state-of-the-art technology to digitally reconstruct the work, preserving the original format whilst repairing imperfections present in the aged copy. In rare cases, an imperfection in the original, such as a blemish or missing page, may be replicated in our edition. We do, however, repair the vast majority of imperfections successfully; any imperfections that remain are intentionally left to preserve the state of such historical works.
In geometric hashing, the collection of models are used in a preprocessing phase (executed off-line and only once) in order to build a hash table data structure. The data structure encodes the information about the models in a highly redundant multiple-viewpoint way. During the recognition phase, when presented with a scene and extracted features, the hash table data structure is used to index geometric properties of the scene features to candidate matching models. A search is still required over features in the scene. However, the geometric hashing scheme no longer requires a search over the features in the model sets. The result is that the recognition phase offers computational efficiencies over more traditional model-based vision methods. As we describe elsewhere there is parallelism available in the search over scene features, and there is parallelism in the indexing process inherent within the hashing scheme. Further, the computational efficiencies afforded by geometric hashing translate into a reduction in the number of independent tasks that may be simultaneously conducted (at the expense of increased storage requirements), thereby decreasing the number of processors that are needed in a parallel implementation.
In this paper, we report on an implementation on the Connection Machine of one of the algorithms described in We also describe a number of modifications that are useful for efficient parallel implementation of the method for the particular case of point features. The last-section of the paper presents the results of a number of experiments that were carried out using this implementation. These experiments had a two-fold purpose: first, we evaluated the real time performance of the implementation with a large database of models; second, the behavior of the method in the case of rigid or similarity transforms was examined, as a function of the noise present in the input. Other statistics are also presented.
About the Publisher
Forgotten Books publishes hundreds of thousands of rare and classic books. Find more at www.forgottenbooks.com
This book is a reproduction of an important historical work. Forgotten Books uses state-of-the-art technology to digitally reconstruct the work, preserving the original format whilst repairing imperfections present in the aged copy. In rare cases, an imperfection in the original, such as a blemish or missing page, may be replicated in our edition. We do, however, repair the vast majority of imperfections successfully; any imperfections that remain are intentionally left to preserve the state of such historical works.
30 pages, Hardcover
First published September 27, 2015
Ratings & Reviews
Friends & Following
Create a free account to discover what your friends think of this book!
Community Reviews
No one has reviewed this book yet.