Intelligent Automated Methods For Monitoring Agricultural Resources
Remote sensing images have been significant sources of information for many applications, specifically for monitoring agricultural and environmental resources. Yet knowledge extraction from these images is often performed by domain experts using heavily interactive computer-aided photo-interpretations due to lack of powerful automated methods, as evidenced by many presentations at the remote-sensing conferences (i.e., IEEE IGARSS, ISPRS Congress, GeoCAP), and numerous papers regularly published at the remote-sensing journals (i.e., IEEE TGARS, ISPRS Journal of Photogr. & Remote Sensing, Computers in Agriculture, IEEE JSTARS, Remote Sensing of Environment), describing their user defined decisions (optimized by user-set image-specific thresholds) for interactive information mining. Improved spatial/spectral resolution in recent years (i.e., less than 50 cm spatial resolution, multi- or even hyper-spectral bands, with reduced revisit time) provides details for precise monitoring, in expense of making the problem even more difficult. This project aims to uniquely propose an automated method (with limited user interaction) based on novel advanced similarity criteria utilizing spectral and spatial characteristics for improved manifold learning techniques and on novel hybrid approach combining pixel-based clustering with parcel-based object analysis for clustering large data sets of very-high resolution remote sensing images. The proposed methods will be primarily used for monitoring agriculture in Europe an Turkey, and in the long run they will be extended for monitoring other natural resources for sustainable development.
Publications
- 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
- Kadim Taşdemir, Berna Yalçin, Isa Yildirim, "Approximate spectral clustering with utilized similarity information using geodesic based hybrid distance measures", Pattern Recognition, Volume 48, Issue 4, Pages 1465-1477, 2015.
- Y. Moazzen, B. Yalçın, K. Taşdemir, "Sampling based approximate spectral clustering ensemble for unsupervised land cover identification", IEEE International Geoscience and Remote Sensing Symposium (IGARSS), 26-31 July 2015, Milan, Italy, 2015.
- E. Pala, K. Taşdemir, D. Koc-San, "Unsupervised extraction of greenhouses using approximate spectral clustering ensemble", IEEE International Geoscience and Remote Sensing Symposium (IGARSS), 26-31 July 2015, Milan, Italy, 2015.
- K.Taşdemir, Y. Moazzen, İ. Yıldırım, “Geodesic based similarities for approximate spectral clustering”, ICPR2014, 24-28 August 2014, Stockholm, Sweden, 2014.
- K. Taşdemir D. Koç-San, “Unsupervised extraction of greenhouses using Worldview-2 images”, IGARSS’2014, 13-18 July 2014, Quebec, Canada, 2014.
- B. Yalçın, Y. Moazzen, K. Taşdemir, "Yaklaşık spektral öbekleme birleşimiyle fındık arazilerinin bulunması", 2nd Workshop on Remote Sensing Signal and Image Processing, IEEE Signal Processing and Communications Applications Conference (SIU), 18-21 Mayıs 2015, Malatya, 2015.
- B. Yalçın, K. Taşdemir, “Büyül veri kümelerinin yaklaşık spectral öbeklemesi için k-ortalama++ nicemleme yönteminin kullanılması”, IEEE Signal Processing and Communications Applications Conference (SIU), 23-25 Nisan 2014, Trabzon, 2014
- Y. Moazzen, K. Taşdemir, “Uzaktan algı görüntülerinin yaklaşık spektral öbeklenmesinde jeodezik bazlı hibriıt benzerlik ölçütleri”, IEEE Signal Processing and Communications Applications Conference (SIU), 23-25 Nisan 2014, Trabzon, 2014
