Alexander Suhre | Bilkent University (original) (raw)
Uploads
Papers by Alexander Suhre
2011 19th European Signal Processing Conference, 2011
Some biomedical images show a large quantity of different junctions and sharp corners. It is poss... more Some biomedical images show a large quantity of different junctions and sharp corners. It is possible to classify several types of biomedical images in a region covariance approach. Cancer cell line images are divided into small blocks and covariance matrices of image blocks are computed. Eigenvalues of the covariance matrices are used as classification parameters in a Bayesian framework using the sparsity of the parameters in a transform domain. The efficiency of the proposed method over classification using standard Support Vector Machines (SVM) is demonstrated on biomedical image data.
Kernel density estimation (KDE) is a popular approach for non-parametric estimation of an underly... more Kernel density estimation (KDE) is a popular approach for non-parametric estimation of an underlying density from data. The performance of KDE is mainly dependent on the bandwidth choice of the very kernel. This article presents various methods of estimating the bandwidth using sparsity in the Fourier transform domain. It uses the Total Variation (TV) and Filtered Variation (FV) cost functions to estimate the bandwidth. Simulation results indicate that, over a set of distributions of interest, the presented approaches are able to outperform classical approaches. Index Terms KDE, Bandwidth, Fourier Domain, Total Variation.
Novel Methods for Microscopic Image Processing, Analysis, Classification and Compression Alexande... more Novel Methods for Microscopic Image Processing, Analysis, Classification and Compression Alexander Suhre Ph.D. in Electrical and Electronics Engineering Supervisor: Prof. Dr. Ahmet Enis Çetin May 2013 Microscopic images are frequently used in medicine and molecular biology. Many interesting image processing problems arise after the initial data acquisition step, since image modalities are manifold. In this thesis, we developed several algorithms in order to handle the critical pipeline of microscopic image storage/compression and analysis/classification more efficiently. The first step in our processing pipeline is image compression. Microscopic images are large in size (e.g. 100K-by-100K pixels), therefore finding efficient ways of compressing such data is necessary for efficient transmission, storage and evaluation. We propose an image compression scheme that uses the color content of a given image, by applying a block-adaptive color transform. Microscopic images of tissues have a...
The invention relates to a method for recognizing shadowing of a sensor device (4) for a driver a... more The invention relates to a method for recognizing shadowing of a sensor device (4) for a driver assistance system (2) of a motor vehicle (1) by an object (8), wherein at least one of the sensor device (4) detected echo signal (S2) having a a detecting region (e) for the sensor means (4) is a distance between the sensor device (4), characterized and the object (8), received by means of a computing device (3), is determined and is checked against the at least one received echo signal (S2), whether the detection area (e) is at least partially shadowed of the sensor device (4) through the object (8), wherein the at least one echo signal (S2) by means of the computing means (3) a discrete distance value (B1, B2, B3) of a plurality of discrete distance values (B1, B2, B3) is assigned for the associated discrete distance value (B1, B2, B3) a power value (P) for the echo signal (S2) is determined and the section z hattung the sensor device (4) is detected if a deviation between the power ...
2018 19th International Conference on Thermal, Mechanical and Multi-Physics Simulation and Experiments in Microelectronics and Microsystems (EuroSimE), 2018
Simulation of sensor readings is important within the field of advanced driver-assistance systems... more Simulation of sensor readings is important within the field of advanced driver-assistance systems, specifically with respect to feasibility studies of high-risk scenarios, where carrying out such test drives in the real world is expensive. Automotive radar signals are multi-dimensional and their data sizes are large, due to multiple measurements being taken within a short time frame in order to resolve ambiguities. Analyzing such signals and generating models from them is therefore no easy task. This paper presents an approach to generate models from data using neural networks. Conditional variational auto-encoders are used for their ability to learn complex distributions and can easily be trained via gradient descent-type algorithms. Our method is able to handle big data sizes and generate synthetic data of high accuracy.
2011 19th European Signal Processing Conference, Aug 1, 2011
2015 16th International Radar Symposium (IRS), 2015
2013 21st Signal Processing and Communications Applications Conference (SIU), 2013
Signal, Image and Video Processing, 2014
Signal, Image and Video Processing, 2014
2010 IEEE 18th Signal Processing and Communications Applications Conference, 2010
Optical Engineering, 2011
AStA Advances in Statistical Analysis, 2013
2013 IEEE International Conference on Acoustics, Speech and Signal Processing, 2013
2011 19th European Signal Processing Conference, 2011
Some biomedical images show a large quantity of different junctions and sharp corners. It is poss... more Some biomedical images show a large quantity of different junctions and sharp corners. It is possible to classify several types of biomedical images in a region covariance approach. Cancer cell line images are divided into small blocks and covariance matrices of image blocks are computed. Eigenvalues of the covariance matrices are used as classification parameters in a Bayesian framework using the sparsity of the parameters in a transform domain. The efficiency of the proposed method over classification using standard Support Vector Machines (SVM) is demonstrated on biomedical image data.
Kernel density estimation (KDE) is a popular approach for non-parametric estimation of an underly... more Kernel density estimation (KDE) is a popular approach for non-parametric estimation of an underlying density from data. The performance of KDE is mainly dependent on the bandwidth choice of the very kernel. This article presents various methods of estimating the bandwidth using sparsity in the Fourier transform domain. It uses the Total Variation (TV) and Filtered Variation (FV) cost functions to estimate the bandwidth. Simulation results indicate that, over a set of distributions of interest, the presented approaches are able to outperform classical approaches. Index Terms KDE, Bandwidth, Fourier Domain, Total Variation.
Novel Methods for Microscopic Image Processing, Analysis, Classification and Compression Alexande... more Novel Methods for Microscopic Image Processing, Analysis, Classification and Compression Alexander Suhre Ph.D. in Electrical and Electronics Engineering Supervisor: Prof. Dr. Ahmet Enis Çetin May 2013 Microscopic images are frequently used in medicine and molecular biology. Many interesting image processing problems arise after the initial data acquisition step, since image modalities are manifold. In this thesis, we developed several algorithms in order to handle the critical pipeline of microscopic image storage/compression and analysis/classification more efficiently. The first step in our processing pipeline is image compression. Microscopic images are large in size (e.g. 100K-by-100K pixels), therefore finding efficient ways of compressing such data is necessary for efficient transmission, storage and evaluation. We propose an image compression scheme that uses the color content of a given image, by applying a block-adaptive color transform. Microscopic images of tissues have a...
The invention relates to a method for recognizing shadowing of a sensor device (4) for a driver a... more The invention relates to a method for recognizing shadowing of a sensor device (4) for a driver assistance system (2) of a motor vehicle (1) by an object (8), wherein at least one of the sensor device (4) detected echo signal (S2) having a a detecting region (e) for the sensor means (4) is a distance between the sensor device (4), characterized and the object (8), received by means of a computing device (3), is determined and is checked against the at least one received echo signal (S2), whether the detection area (e) is at least partially shadowed of the sensor device (4) through the object (8), wherein the at least one echo signal (S2) by means of the computing means (3) a discrete distance value (B1, B2, B3) of a plurality of discrete distance values (B1, B2, B3) is assigned for the associated discrete distance value (B1, B2, B3) a power value (P) for the echo signal (S2) is determined and the section z hattung the sensor device (4) is detected if a deviation between the power ...
2018 19th International Conference on Thermal, Mechanical and Multi-Physics Simulation and Experiments in Microelectronics and Microsystems (EuroSimE), 2018
Simulation of sensor readings is important within the field of advanced driver-assistance systems... more Simulation of sensor readings is important within the field of advanced driver-assistance systems, specifically with respect to feasibility studies of high-risk scenarios, where carrying out such test drives in the real world is expensive. Automotive radar signals are multi-dimensional and their data sizes are large, due to multiple measurements being taken within a short time frame in order to resolve ambiguities. Analyzing such signals and generating models from them is therefore no easy task. This paper presents an approach to generate models from data using neural networks. Conditional variational auto-encoders are used for their ability to learn complex distributions and can easily be trained via gradient descent-type algorithms. Our method is able to handle big data sizes and generate synthetic data of high accuracy.
2011 19th European Signal Processing Conference, Aug 1, 2011
2015 16th International Radar Symposium (IRS), 2015
2013 21st Signal Processing and Communications Applications Conference (SIU), 2013
Signal, Image and Video Processing, 2014
Signal, Image and Video Processing, 2014
2010 IEEE 18th Signal Processing and Communications Applications Conference, 2010
Optical Engineering, 2011
AStA Advances in Statistical Analysis, 2013
2013 IEEE International Conference on Acoustics, Speech and Signal Processing, 2013