Please use this identifier to cite or link to this item: http://scholarbank.nus.edu.sg/handle/10635/126065
Title: THE ROLE OF TEXTURE IN INDOOR SCENE RECOGNITION
Authors: SHAHZOR AHMAD
Keywords: homogeneous texture, affine rectification, projective-invariant detection, indoor geometric layout, indoor scene recognition, deep ConvNets
Issue Date: 13-Jan-2016
Abstract: THIS THESIS ADVANCES A NOVEL PARADIGM INVOLVING THE USE OF PLANAR HOMOGENEOUS TEXTURE FOR AN IMPROVED INDOOR SCENE REPRESENTATION AND SUBSEQUENT CLASSIFICATION. SUCH TEXTURE MANIFESTS IN THE FORM OF REGULARLY REPEATING STRUCTURAL OR ARCHITECTURAL ELEMENTS, OR AS UNIFORM PRINTED OR ENGRAVED PATTERNS ON MATERIAL, AND IS ABUNDANTLY PRESENT IN MAN-MADE ENVIRONMENTS. IN ORDER TO MITIGATE IN-CLASS VARIATION ARISING OUT OF VIEWPOINT DIFFERENCES AND PERSPECTIVE PROJECTION IN IMAGES, THE PROBLEM OF PLANAR RECTIFICATION OF HOMOGENEOUS TEXTURE IS FIRST ADDRESSED. A TEXTURE FREQUENCY PROJECTION MODEL IS DEVELOPED IN ORDER TO RECOVER PLANE PROJECTIVE PARAMETERS, ALLOWING AN AFFINE-AMBIGUOUS RECTIFICATION. ROBUST RECOVERY OF OPTIMAL DOMINANT INSTANTANEOUS FREQUENCY IS DEMONSTRATED IN IMAGES MARRED WITH OUTLIERS, CLUTTER AND PHOTOMETRIC SEVERITIES. THE APPROACH MAKES FOR A HAND-CRAFTED METHOD FOR THE LOCALIZATION OF SEMANTICALLY MEANINGFUL, POTENTIALLY LARGE TEXTURED REGIONS, AS OPPOSED TO LEARNIN
URI: http://scholarbank.nus.edu.sg/handle/10635/126065
Appears in Collections:Ph.D Theses (Open)

Show full item record
Files in This Item:
File Description SizeFormatAccess SettingsVersion 
AhmadS.pdf22.64 MBAdobe PDF

OPEN

NoneView/Download

Page view(s)

163
checked on Nov 9, 2018

Download(s)

66
checked on Nov 9, 2018

Google ScholarTM

Check


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