2012 International Conference on Computational Intelligence and Software Engineering (CiSE 2012)(E-BOOK)

Wuhan,China,2012-12-142012-12-162012

ISBN: 978-1-61896-036-8 Scientific Research Publishing

E-Book 275pp Pub. Date: December 2012

Category: Computer Science & Communications

Price: $100

Title: A Visual Words Simplified Mehtod for Fast Pedestrian Detection
Source: 2012 International Conference on Computational Intelligence and Software Engineering (CiSE 2012)(E-BOOK) (pp 47-50)
Author(s): Xingguo Zhang, Graduate School of Systems Science and Technology,Akita Prefectural University,Yurihonjo City, Akita, Japan
Kazuki Saruta, Department of Electronics and Information Systems,Akita Prefectural University,Yurihonjo City, Akita, Japan
Yuki Terata, Department of Electronics and Information Systems,Akita Prefectural University,Yurihonjo City, Akita, Japan
Guoyue Chen, Department of Electronics and Information Systems,Akita Prefectural University,Yurihonjo City, Akita, Japan
Abstract: Pedestrian detection is an important area in computer vision with key applications in intelligent vehicles. The objective of this paper is to provide a method to simplify the visual words based on Bag-of-Features (BoF) algorithm for fast pedestrian detection in low resolution infrared images. A dense-SIFT descriptor is adopted to extract the local features of image and construct a visual vocabulary. We proposed a new approach to remove some redundant visual words which are useless for classification, as a result, the time of the classifier to train and recognize image samples will be remarkably reduced. The experiment shows that BoF is an effective method to detect the image of pedestrian with position shift, and this method can strikingly speed up the learning and recognizing process with only a slight decrease in recognition accuracy.
Free SCIRP Newsletters
Copyright © 2006-2024 Scientific Research Publishing Inc. All Rights Reserved.
Top