The list of projects enrolled;“Automatic target recognition system on SAR images”Analysis of satellite SAR images to recognize and classify military and civil targets.A complete target detection/recognition system is developed for the military customer. Design of algorithms from preprocessing to recognition of targets are covered. The filters and recognition algorithms (including detection, segmentation, and classification) are designed at Matlab environment and the implementation is done at C++ environment with the use of OpenMP, CUDA utilities. As a new approach to the land cover classification problem, deep learning approaches are being tried. “Analysis and classification of hyperspectral data for target recognition”Analysis of Hyperspectral data generated by airborne and ground platforms for the aim of detecting military targets. Matlab environment is used for design and testing of the algorithms.“Analysis of Lidar data for Forest Management”A research project which includes the design and implementation of the Lidar data for the aim of detailed analysis of the forest area. “SAR Feature Toolbox for Erdas Imagine”Design and development of SAR image processing algorithms as a toolbox for the commercial GIS software Imagine.“All condition Operations and Innovative Cockpit Infrastructure”The project is supported by European Commission in framework of FP7. We have developed algorithms for enhancing the images on vision systems to increase situational awareness of the pilot, to support landing decision with an ultimate goal of decreasing weather dependent delays. Images from different sensors (EO, IR, MMWR, SAR) are processed by image processing algorithms and fused at different levels to feed outputs to cockpit display systems. Two versions of the system (rotary wing and fixed wing) are integrated to Agusta Westland UK and Thales FR test beds successfully.

Courses given;Undergraduate: Software engineering, database management systems, computer networksGraduate: Pattern recognition, machine learning

Technical management of a research project supported by TÜBİTAK: “OAGAIT: A Decision Support System for Grading Knee Osteoarthritis Using Gait Data” (Ph.D. thesis). Analysis of human gait patterns to assist physicians for decision making about the grading of the neuromusculosceletal disorders. Many kinds of neural networks (like MLP, time-delay NN, convolutional NN, and recurrent NN etc.) are implemented at Matlab environment. “Combining classifiers” and “mixture of experts” methods where different classifiers (decision trees, SVM etc.) are implemented and the results are published.