An Approximate Spectral Clustering Ensemble for High Spatial Resolution Remote-Sensing Images

Unsupervised clustering of high spatial resolution remote-sensing images plays a significant role in detailed land cover identification, especially for agricultural and environmental monitoring. A recently promising method is approximate spectral clustering (SC) which enables spectral partitioning for large datasets to extract clusters with distinct characteristics without a parametric model. It also facilitates the use of various information types via advanced similarity criteria. However, it requires an empirical selection of a similarity criterion optimal for the corresponding application. To address this challenge, we propose an approximate SC ensemble (ASCE2) which fuses partitionings obtained by different similarity representations. Contrary to existing spectral ensembles for remote-sensing applications, the proposed ASCE2 employs neural gas quantization instead of random sampling, advanced similarity criteria instead of traditional distance-based Gaussian kernel with different decay parameters, and a two-level ensemble. We evaluate the proposed ASCE2 with three measures (accuracy, adjusted Rand index, and normalized mutual information) using five remote-sensing images, two of which are commonly available. We apply the ASCE2 in two applications for agricultural monitoring: 1) land-cover identification to determine orchard fields using a WorldView-2 image (0.5-m spatial resolution) and 2) finding lands in good agricultural condition usingmultitemporal RapidEye images (5-mspatial resolution). Experimental results indicate a significant betterment of the resulting partitionings obtained by the proposed ensemble, with respect to the evaluation measures in these applications.

Kadim Taşdemir, Yaser Moazzen, Isa Yıldırım, "An Approximate Spectral Clustering Ensemble for High Spatial Resolution Remote-Sensing Images", IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, DOI: 10.1109/JSTARS.2015.2424292, 2015.

Approximate spectral clustering with utilized similarity information using geodesic based hybrid distance measures

Spectral clustering has been popular thanks to its ability to extract clusters of varying characteristics without using a parametric model in expense of high computational cost required for eigendecomposition of pairwise similarities. In order to utilize its advantages in large datasets where it is infeasible due to its computational burden, approximate spectral clustering (ASC) methods apply spectral clustering on a reduced set of points (data representatives) selected by sampling or quantization. This two-step approach (i.e. finding the representatives and then clustering them) brings new opportunities for precise similarity definition such as manifold based topological relations, data distribution within the Voronoi polyhedra of the representatives, and their geodesic distance information, which are often ignored in similarity definition for ASC. In this study, we propose geodesic based hybrid similarity criteria which enable the use of different types of information for accurate similarity representation in ASC. Despite the fact that geodesic concept has been widely used in clustering, our contribution is the unique way of representing data topology to form geodesic relations and jointly harnessing various information types including topology, distance and density.The proposed criteria are tested using both sampling (selective sampling) and quantization (neural gas and k-means) approaches.Experiments on artificial datasets, well-known small/medium-size real datasets, and four large datasets (four remote-sensing images), with different types of clusters, show that the proposed geodesic based hybrid similarity criteria outperform traditional similarity criteria in terms of clustering accuracies and several cluster validity indices.

K.Taşdemir, B. Yalçın, İ. Yıldırım, "Approximate spectral clustering with utilized similarity information using geodesic based hybrid distance measures", Pattern Recognition (2014), http://dx.doi.org/10.1016/j.patcog.2014.10.023

A hybrid method combining SOM-based clustering and object-based analysis for identifying land in good agricultural condition

Remotely sensed imagery is currently used as an efficient tool for agricultural management and monitoring. In addition, the use of remotely sensed imagery in Europe has been extended towards determination of the areas potentially eligible for the farmer subsidies under the Common Agricultural Policy (CAP), through interactive or automatic land cover identification. For accurate quantification and fast identification of agricultural land cover areas from the imagery, a hybrid method, which combines automated clustering of self-organizing maps with object based image analysis, and called SOM + OBIA, is proposed. Performance analysis on three test zones (using multi-temporal Rapideye imagery) indicates that for the basic land cover categories (forest, water, vegetated areas, bare areas and sealed surfaces), unsupervised classification with the proposed SOM + OBIA method achieves an identification accuracy comparable to the accuracy of the traditional interactive object oriented analysis, with considerably less user interaction.

K. Taşdemir, P. Milenov and B. Tapsall, “A hybrid method combining SOM-based clustering and object-based analysis for identifying land in good agricultural condition,” Computers and Electronics in Agriculture, 83, 92-101, 2012.

Vector quantization based approximate spectral clustering

Spectral partitioning, recently popular for unsupervised clustering, is infeasible for large datasets due to its computational complexity and memory requirement.Therefore,approximate spectral clustering of data representatives (selected by various sampling methods) was used. Alternatively,we propose to use neural networks (self-organizing maps and neural gas),which are shown successful in quantization with small distortion, as preliminary sampling for approximate spectral clustering (ASC).We show that they usually outperform k-means sampling (which was shown superior to various sampling methods), in terms of clustering accuracy obtained by ASC. More importantly, for quantization based ASC, we introduce a local density-based similarity measure – constructed without any user-set parameter – which achieves accuracies superior to the accuracies of commonly used distance based similarity.

K. Taşdemir, “Vector quantization based approximate spectral clustering of large datasets,” Pattern Recognition, 45 (8), 3034-3044 , 2012.

About me

Kadim Tasdemir received the B.S. degree in Electrical and Electronics Engineering from Bogazici University, Istanbul, Turkey, the M.S. degree in Computer Engineering from Istanbul Technical University, Istanbul, Turkey, and the Ph.D. degree in Electrical and Computer Engineering from Rice University, Houston, TX, USA, in 2001, 2004, and 2008, respectively.

Dr. Tasdemir's research interests include detailed knowledge discovery from high-dimensional and large data sets (especially remote sensing images) using machine learning (self-organized learning in particular), data mining and pattern recognition. He is a recepient of FP7 Marie Curie Career Integration Grant and TUBITAK Career Grant. He is a principal investigator for a joint project funded by TUBITAK for food quality assessment based on hyperspectral image analysis.

Before joining AIU, he was a researcher at the European Commission Joint Research Centre (JRC), (Institute for Environment and Sustainability) from 2009 to 2012, where he worked on automated control methods for monitoring agricultural resources using remote sensing imagery. Based on his research excellence and contribution, he received 2011 IES Best Young Scientist Award. Before JRC, he worked as Assistant Professor at Department of Computer Engineering, Yasar University, Izmir, Turkey, in 2008-2009. During 2003-2008, he was a research assistant at Rice University, where he developed visualization and clustering methods using neural computation for detailed knowledge discovery, sponsored by NASA Applied Information Systems Research Program. He was also awarded “Rice University Robert Patten Award” for his contributions to graduate life. During 2001-2003, he was a research assistant at Istanbul Technical University, where he worked on License Plate Recognition project.

Dr. Tasdemir is a member of IEEE, IEEE Computational Intelligence Society, IEEE Geoscience and Remote Sensing Society, IAPR- TC7 Remote Sensing and Mapping, and IAPR-TC15 Graph Based Representations. He is the founding chair of IEEE GRSS Turkey Chapter since December 2012. Despite being a new chapter with few members, Turkey Chapter received the GRSS Chapter Excellence Award based on its activities in 2014. He is one of the organizers of Remote Sensing Summer School in TUBITAK BILGEM in 2014. He is a co-chair of National Workshop on Remote Sensing Signal and Image Processing since 2014. He is also a program committee member for Workshop on Self-Organizing Maps, Recent Advances in Space Technologies.

Dr. Tasdemir is on Editorial Board for ISPRS Journal of Photogrammetry and Remote Sensing. He is also a reviewer for several journals including IEEE Trans. on Neural Networks and Learning Systems, IEEE Trans. on Image Processing, IEEE TGARS, IEEE JSTARS, IEEE Trans. on Cybernetics, IEEE Signal Processing Letters, International Journal of Remote Sensing, Neural Processing Letters.