Texture segmentation using the mixtures of principal component analyzers

Date

2003-01-01

Author

Musa, MEM
Duin, RPW
de Ridder, D
Atalay, Mehmet Volkan

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

114
views

0
downloads

The problem of segmenting an image into several modalities representing different textures can be modelled using Gaussian mixtures. Moreover, texture image patches when translated, rotated or scaled lie in low dimensional subspaces of the high-dimensional space spanned by the grey values. These two aspects make the mixture of local subspace models worth consideration for segmenting this type of images. In recent years a number of mixtures of local PCA models have been proposed. Most of these models require the user to set the number of subspaces and subspace dimensionalities. To make the model autonomous, we propose a greedy EM algorithm to find a suboptimal number of subspaces, besides using a global retained variance ratio to estimate for each subspace the dimensionality that retains the given variability ratio. We provide experimental results for testing the proposed method on texture segmentation.

Subject Keywords

Models

URI

https://hdl.handle.net/11511/55762

Journal

COMPUTER AND INFORMATION SCIENCES - ISCIS 2003

Collections

Department of Computer Engineering, Article

Suggestions

OpenMETU
Core

2d polynomial interpolation: A symbolic approach with mathematica Yazıcı, Adnan; Ergenc, T (2005-01-01) This paper extends a previous work done by the same authors on teaching 1d polynomial interpolation using Mathematica [1] to higher dimensions. In this work, it is intended to simplify the the theoretical discussions in presenting multidimensional interpolation in a classroom environment by employing Mathematica's symbolic properties. In addition to symbolic derivations, some numerical tests are provided to show the interesting properties of the higher dimensional interpolation problem. Runge's phenomenon w...
Color texture morphology Tari, S (1999-06-23) The paper presents a computational model to extract local symmetry axes corresponding to the areas of homogeneous texture in an image. The approach is to construct an energy functional whose minimizer is a special distance function which attains its maxima at the boundaries between different texture regions and decays away from texture boundaries. Local symmetry axes are extracted from this distance function. The method extends to color texture images. Texture homogeneity is measured using the filter respon...
Texture and edge preserving multiframe super-resolution Turgay, Emre; Akar, Gözde (Institution of Engineering and Technology (IET), 2014-09-01) Super-resolution (SR) image reconstruction refers to methods where a higher resolution image is reconstructed using a set of overlapping aliased low-resolution observations of the same scene. Although edge preservation has been a widely explored topic in SR literature, texture-specific regularisation has recently gained interest. In this study, texture-specific regularisation is handled as a post-processing step. A two stage method is proposed, comprising multiple SR reconstructions with different regularis...
3B Izometrik Şekil Eşleme Sahillioğlu, Yusuf (2010-06-01) 3B izometrik şekiller arasındaki eşleme problemini ele alıyoruz. Önerdiğimiz yöntem, verilen iki izometrik şekil arasındaki izometrik sapmayı enküçülten optimal eşlemeyi otomatik olarak bulabilmektedir.İzometri hatasını iki adımda eniyiliyoruz. İlk adımda, şekil yüzeylerinden örneklenmiş bir örnek 3B noktalar kesel ilginlik bilgisine dayanarak spektral uzaya aktarılır.İlk eşleme spektral uzayda tam iki kısımlı bir çizge eşleştirme yöntemi kullanarak izometri hatasının polinom zamanda enküçültülmesiyle eld...
Semi-automated registration of BILSAT-1 XS imagery via global orthorectified landsat ETM plus data set Ok, Ali Oezguen (2007-06-16) In this study, a semi-automated registration methodology that can be adapted to BILSAT-1 XS images is proposed. Orthorectified and geodetically accurate global land datasets generated from the Landsat ETM+ images are used as the reference sources. Three different test sites are employed to test the accuracies of the registration. The accuracies of the registration were assessed using Independent Check Points (ICPs) which were manually collected from 1:25000-scale digital topographic maps of Turkey. The comp...

Citation Formats

M. Musa, R. Duin, D. de Ridder, and M. V. Atalay, “Texture segmentation using the mixtures of principal component analyzers,” COMPUTER AND INFORMATION SCIENCES - ISCIS 2003, pp. 505–512, 2003, Accessed: 00, 2020. [Online]. Available: https://hdl.handle.net/11511/55762.