Texture Image Analysis for Larger Lattice Structure Using Orthogonal Polynomial Frameworkopen access
- Authors
- Ganesan, L.; Umarani, C.; Kaliappan, M.; Vimal, S.; Kadry, Seifedine; Nam, Yunyoung
- Issue Date
- Nov-2022
- Publisher
- Kaunas University of Technology
- Keywords
- Texture Representation; Polynomials; Statistical Tests; Local and Global descriptors; Standard Texture Images; Classification
- Citation
- Information Technology and Control, v.51, no.3, pp 531 - 544
- Pages
- 14
- Journal Title
- Information Technology and Control
- Volume
- 51
- Number
- 3
- Start Page
- 531
- End Page
- 544
- URI
- https://scholarworks.bwise.kr/sch/handle/2021.sw.sch/21851
- DOI
- 10.5755/j01.itc.51.3.29322
- ISSN
- 1392-124X
2335-884X
- Abstract
- An Orthogonal Polynomial Framework using 5 x5 mathematical model is proposed and attempted for the texture image analysis. The Orthogonal Polynomial Framework has been shown effective for image with larger image grid size of (5 x 5) or (7 x 7) or (9*9) or even higher, to analyse textured surfaces. The image region (5 x 5) under consideration is evaluated to be textured or untextured using a statistical approach. The textured region is represented locally as well as globally by suitable descriptors for further analysis. The local descriptor is termed as pro5num and histogram of these pro5nums is called the pro5spectrum, the global descriptor. The novelty of this scheme is its representation of texture after it identifies the presence and the model can be extended to any image size region. This method works fine for standard database of texture images. The texture images have been shown successfully represented by the descriptors. The proposed scheme has been successfully applied for the supervised texture classification and the classification accuracies are comparable with recent results. Further texture analysis problems using these descriptors are in progress.
- Files in This Item
- There are no files associated with this item.
- Appears in
Collections - College of Engineering > Department of Computer Science and Engineering > 1. Journal Articles
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.