Please use this identifier to cite or link to this item: http://scholarbank.nus.edu.sg/handle/10635/121995
Title: IMAGE CLASSIFICATION USING INVARIANT LOCAL FEATURES AND CONTEXTUAL INFORMATION
Authors: RAMESH BHARATH
Keywords: Object Classification, Log-Polar Transform, Contextual Information, Feature Extraction, Bag of words model, Image Classification
Issue Date: 18-Aug-2015
Source: RAMESH BHARATH (2015-08-18). IMAGE CLASSIFICATION USING INVARIANT LOCAL FEATURES AND CONTEXTUAL INFORMATION. ScholarBank@NUS Repository.
Abstract: Despite many successful applications, computers have made little progress in generalizing object appearance even under controlled sensing environments, whereas humans effortlessly categorize myriad objects by utilizing a variety of visual cues. Taking inspiration from the human vision system, the central theme of this thesis is a cue-based approach to object categorization. The first contribution is a novel local shape descriptor, robust to scale and rotation variations, using a biologically inspired technique called log-polar transform. The second contribution is a novel fusion of grayscale and binary shape cues for classification of object images from the ETH-80 dataset. Then, we propose a generic cue-based image classification framework with excellent performance compared to existing works on widely used datasets, such as Caltech-101 and Flickr-101. Lastly, we present a novel real-world traffic monitoring application using log-polar transform for encoding multiple depth-of-field information, and thus enabling a longer tracking range for reliable over-speeding vehicle detection.
URI: http://scholarbank.nus.edu.sg/handle/10635/121995
Appears in Collections:Ph.D Theses (Open)

Show full item record
Files in This Item:
File Description SizeFormatAccess SettingsVersion 
RameshB.pdf5.59 MBAdobe PDF

OPEN

NoneView/Download

Page view(s)

109
checked on Jan 18, 2018

Download(s)

167
checked on Jan 18, 2018

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.