Zeynel Deprem - Academia.edu (original) (raw)

Papers by Zeynel Deprem

Signal representation in Time-Frequency (TF) domain is valuable in many applications including ra... more Signal representation in Time-Frequency (TF) domain is valuable in many applications including radar imaging and inverse synthetic aparture radar. TF representation allows us to identify signal components or features in a mixed time and frequency plane. There are several well-known tools, such as Wigner-Ville Distribution (WVD), Short-Time Fourier Transform (STFT) and various other variants for such a purpose. The main requirement for a TF representation tool is to give a highresolution view of the signal such that the signal components or features are identifiable. A commonly used method is the reassignment process which reduces the cross-terms by artificially moving smoothed WVD values from their actual location to the center of the gravity for that region. In this article, we propose a novel reassignment method using the Conditional Generative Adversarial Network (CGAN). We train a CGAN to perform the reassignment process. Through examples, it is shown that the method generates h...

SPARSITY AND CONVEX PROGRAMMING IN TIME-FREQUENCY PROCESSING Zeynel Deprem Ph.D. in Electrical an... more SPARSITY AND CONVEX PROGRAMMING IN TIME-FREQUENCY PROCESSING Zeynel Deprem Ph.D. in Electrical and Electronics Engineering Advisor: Prof. Dr. Ahmet Enis Çetin December, 2014 In this thesis sparsity and convex programming-based methods for timefrequency (TF) processing are developed. The proposed methods aim to obtain high resolution and cross-term free TF representations using sparsity and lifted projections. A crucial aspect of Time-Frequency (TF) analysis is the identification of separate components in a multi component signal. Wigner-Ville distribution is the classical tool for representing such signals but suffers from cross-terms. Other methods that are members of Cohen’s class distributions also aim to remove the cross terms by masking the Ambiguity Function (AF) but they result in reduced resolution. Most practical signals with time-varying frequency content are in the form of weighted trajectories on the TF plane and many others are sparse in nature. Therefore the problem ca...

2015 23nd Signal Processing and Communications Applications Conference (SIU), 2015

Özetçe-Bu dokümanda çözünürlügü yüksek ve çapraz terim içermeyen Cohen sınıfı bir Zaman-frekans (... more Özetçe-Bu dokümanda çözünürlügü yüksek ve çapraz terim içermeyen Cohen sınıfı bir Zaman-frekans (ZF) dagılımının, çekirdek kestirim yöntemi ile elde edilmesi tanıtılmaktadır. Çekirdek kestirimi, başlangıç taslak bir zaman-frekans dagılımının l1 normuna ait epigraf kümesi üzerine izdüşümü ile elde edilmektedir. Kestirilen çekirdek, sinyalin belirsizlik (Ambiguity) düzlemindeki hizalanması ile uymlu ve çapraz terimleri içermeyecek bir filtreleme saglamaktadır. Anahtar Kelimeler-zaman-frekans, Cohen sınıfı,İzdüşüm.

Z. Deprem, K. Leblebicioglu, O. Arikan, and A. E. C etin, \A Complexity- Reduced ML Parametric Signal Reconstruction Method," EURASIP Journal on Advances in Signal Processing, pp. 1-14, 2011

EURASIP Journal on Advances in Signal Processing

The problem of component estimation fromamulticomponent signal in additive white Gaussian noise i... more The problem of component estimation fromamulticomponent signal in additive white Gaussian noise is considered. A parametric ML approach, where all components are represented as a multiplication of a polynomial amplitude and polynomial phase term, is used. The formulated optimization problem is solved via nonlinear iterative techniques and the amplitude and phase parameters for all components are reconstructed. The initial amplitude and the phase parameters are obtained via time-frequency techniques. An alternative method, which iterates amplitude and phase parameters separately, is proposed. The proposed method reduces the computational complexity and convergence time significantly. Furthermore, by using the proposed method together with ExpectationMaximization (EM) approach, better reconstruction error level is obtained at low SNR. Though the proposed method reduces the computations significantly, it does not guarantee global optimum. As is known, these types of non-linear optimiza...

IEEE Transactions on Aerospace and Electronic Systems, 2015

A crucial aspect of time-frequency (TF) analysis is the identification of separate components in ... more A crucial aspect of time-frequency (TF) analysis is the identification of separate components in a multicomponent signal. The Wigner-Ville distribution is the classical tool for representing such signals, but it suffers from cross-terms. Other methods, which are members of Cohen's class of distributions, also aim to remove the cross-terms by masking the ambiguity function (AF), but they result in reduced resolution. Most practical time-varying signals are in the form of weighted trajectories on the TF plane, and many others are sparse in nature. Therefore, in recent studies the problem is cast as TF distribution reconstruction using a subset of AF domain coefficients and sparsity assumption. Sparsity can be achieved by constraining or minimizing the l 1 norm. In this article, an l 1 minimization approach based on projections onto convex sets is proposed to obtain a high-resolution, cross-term-free TF distribution for a given signal. The new method does not require any parameter adjustment to obtain a solution. Experimental results are presented.

2013 IEEE Global Conference on Signal and Information Processing, 2013

Signal and image reconstruction from Fourier Transform magnitude is a difficult inverse problem. ... more Signal and image reconstruction from Fourier Transform magnitude is a difficult inverse problem. Fourier transform magnitude can be measured in many practical applications, but the phase may not be measured. Since the autocorrelation of an image or a signal can be expressed as convolution of with , it is possible to formulate the inverse problem as a non-negative matrix factorization problem. In this paper, we propose a new algorithm based on the sparse non-negative matrix factorization (NNMF) to estimate the phase of a signal or an image in an iterative manner. Experimental reconstruction results are presented. I.

I. INTRODUCTION IGNALS with time-varying frequency content are encountered in many areas such as ... more I. INTRODUCTION IGNALS with time-varying frequency content are encountered in many areas such as AM/FM communication, radar, sonar applications, medicine (EEG), gravitational analysis, speech, and audio. An important aspect of Time-Frequency (TF) analysis is the identification of separate components in a multi component signal. Highresolution time-frequency (TF) representations and instantaneous frequency (IF)-based methods are needed for analysis, detection and classification of these type signals. TF signal representations enable separation of time-varying components overlapping both in time and frequency domains. It may not be possible to isolate some signal components in one domain using ordinary frequency domain filtering. An important application area of high resolution TF analysis is the radar imaging [1], [2]. The image of the target is obtained from spatial distribution of the reflected signal.

Signal, Image and Video Processing, 2013

Amplitude and phase estimation of AM/FM signals with parametric polynomial representation require... more Amplitude and phase estimation of AM/FM signals with parametric polynomial representation require the polynomial orders for phase and amplitude to be known. But in reality, they are not known and have to be estimated. A well-known method for estimation is the higher-order ambiguity function (HAF) or its variants. But the HAF method has several reported drawbacks such as error propagation and slowly varying or even constant amplitude assumption. Especially for the long duration time-varying signals like AM/FM signals, which require high orders for the phase and amplitude, computational load is very heavy due to nonlinear optimization involving many variables. This paper utilizes a micro-segmentation approach where the length of segment is selected such that the amplitude and instantaneous frequency (IF) is constant over the segment. With this selection first, the amplitude and phase estimates for each micro-segment are obtained optimally in the LS sense, and then, these estimates are concatenated to obtain the overall amplitude and phase estimates. The initial estimates are not optimal but sufficiently close to the optimal solution for subsequent processing. Therefore, by using the initial estimates, the overall polynomial orders for the amplitude and phase are estimated. Using estimated orders, the initial amplitude and phase functions are fitted to the polynomials to obtain the final signal. The method does not use any multivariable nonlinear optimization and is efficient in the sense that the Z. Deprem (B) • A

EURASIP Journal on Advances in Signal Processing, 2011

The problem of component estimation from a multicomponent signal in additive white Gaussian noise... more The problem of component estimation from a multicomponent signal in additive white Gaussian noise is considered. A parametric ML approach, where all components are represented as a multiplication of a polynomial amplitude and polynomial phase term, is used. The formulated optimization problem is solved via nonlinear iterative techniques and the amplitude and phase parameters for all components are reconstructed. The initial amplitude and the phase parameters are obtained via time-frequency techniques. An alternative method, which iterates amplitude and phase parameters separately, is proposed. The proposed method reduces the computational complexity and convergence time significantly. Furthermore, by using the proposed method together with Expectation Maximization (EM) approach, better reconstruction error level is obtained at low SNR. Though the proposed method reduces the computations significantly, it does not guarantee global optimum. As is known, these types of non-linear optimization algorithms converge to local minimum and do not guarantee global optimum. The global optimum is initialization dependent.

Signal representation in Time-Frequency (TF) domain is valuable in many applications including ra... more Signal representation in Time-Frequency (TF) domain is valuable in many applications including radar imaging and inverse synthetic aparture radar. TF representation allows us to identify signal components or features in a mixed time and frequency plane. There are several well-known tools, such as Wigner-Ville Distribution (WVD), Short-Time Fourier Transform (STFT) and various other variants for such a purpose. The main requirement for a TF representation tool is to give a highresolution view of the signal such that the signal components or features are identifiable. A commonly used method is the reassignment process which reduces the cross-terms by artificially moving smoothed WVD values from their actual location to the center of the gravity for that region. In this article, we propose a novel reassignment method using the Conditional Generative Adversarial Network (CGAN). We train a CGAN to perform the reassignment process. Through examples, it is shown that the method generates h...

SPARSITY AND CONVEX PROGRAMMING IN TIME-FREQUENCY PROCESSING Zeynel Deprem Ph.D. in Electrical an... more SPARSITY AND CONVEX PROGRAMMING IN TIME-FREQUENCY PROCESSING Zeynel Deprem Ph.D. in Electrical and Electronics Engineering Advisor: Prof. Dr. Ahmet Enis Çetin December, 2014 In this thesis sparsity and convex programming-based methods for timefrequency (TF) processing are developed. The proposed methods aim to obtain high resolution and cross-term free TF representations using sparsity and lifted projections. A crucial aspect of Time-Frequency (TF) analysis is the identification of separate components in a multi component signal. Wigner-Ville distribution is the classical tool for representing such signals but suffers from cross-terms. Other methods that are members of Cohen’s class distributions also aim to remove the cross terms by masking the Ambiguity Function (AF) but they result in reduced resolution. Most practical signals with time-varying frequency content are in the form of weighted trajectories on the TF plane and many others are sparse in nature. Therefore the problem ca...

2015 23nd Signal Processing and Communications Applications Conference (SIU), 2015

Özetçe-Bu dokümanda çözünürlügü yüksek ve çapraz terim içermeyen Cohen sınıfı bir Zaman-frekans (... more Özetçe-Bu dokümanda çözünürlügü yüksek ve çapraz terim içermeyen Cohen sınıfı bir Zaman-frekans (ZF) dagılımının, çekirdek kestirim yöntemi ile elde edilmesi tanıtılmaktadır. Çekirdek kestirimi, başlangıç taslak bir zaman-frekans dagılımının l1 normuna ait epigraf kümesi üzerine izdüşümü ile elde edilmektedir. Kestirilen çekirdek, sinyalin belirsizlik (Ambiguity) düzlemindeki hizalanması ile uymlu ve çapraz terimleri içermeyecek bir filtreleme saglamaktadır. Anahtar Kelimeler-zaman-frekans, Cohen sınıfı,İzdüşüm.

Z. Deprem, K. Leblebicioglu, O. Arikan, and A. E. C etin, \A Complexity- Reduced ML Parametric Signal Reconstruction Method," EURASIP Journal on Advances in Signal Processing, pp. 1-14, 2011

EURASIP Journal on Advances in Signal Processing

The problem of component estimation fromamulticomponent signal in additive white Gaussian noise i... more The problem of component estimation fromamulticomponent signal in additive white Gaussian noise is considered. A parametric ML approach, where all components are represented as a multiplication of a polynomial amplitude and polynomial phase term, is used. The formulated optimization problem is solved via nonlinear iterative techniques and the amplitude and phase parameters for all components are reconstructed. The initial amplitude and the phase parameters are obtained via time-frequency techniques. An alternative method, which iterates amplitude and phase parameters separately, is proposed. The proposed method reduces the computational complexity and convergence time significantly. Furthermore, by using the proposed method together with ExpectationMaximization (EM) approach, better reconstruction error level is obtained at low SNR. Though the proposed method reduces the computations significantly, it does not guarantee global optimum. As is known, these types of non-linear optimiza...

IEEE Transactions on Aerospace and Electronic Systems, 2015

A crucial aspect of time-frequency (TF) analysis is the identification of separate components in ... more A crucial aspect of time-frequency (TF) analysis is the identification of separate components in a multicomponent signal. The Wigner-Ville distribution is the classical tool for representing such signals, but it suffers from cross-terms. Other methods, which are members of Cohen's class of distributions, also aim to remove the cross-terms by masking the ambiguity function (AF), but they result in reduced resolution. Most practical time-varying signals are in the form of weighted trajectories on the TF plane, and many others are sparse in nature. Therefore, in recent studies the problem is cast as TF distribution reconstruction using a subset of AF domain coefficients and sparsity assumption. Sparsity can be achieved by constraining or minimizing the l 1 norm. In this article, an l 1 minimization approach based on projections onto convex sets is proposed to obtain a high-resolution, cross-term-free TF distribution for a given signal. The new method does not require any parameter adjustment to obtain a solution. Experimental results are presented.

2013 IEEE Global Conference on Signal and Information Processing, 2013

Signal and image reconstruction from Fourier Transform magnitude is a difficult inverse problem. ... more Signal and image reconstruction from Fourier Transform magnitude is a difficult inverse problem. Fourier transform magnitude can be measured in many practical applications, but the phase may not be measured. Since the autocorrelation of an image or a signal can be expressed as convolution of with , it is possible to formulate the inverse problem as a non-negative matrix factorization problem. In this paper, we propose a new algorithm based on the sparse non-negative matrix factorization (NNMF) to estimate the phase of a signal or an image in an iterative manner. Experimental reconstruction results are presented. I.

I. INTRODUCTION IGNALS with time-varying frequency content are encountered in many areas such as ... more I. INTRODUCTION IGNALS with time-varying frequency content are encountered in many areas such as AM/FM communication, radar, sonar applications, medicine (EEG), gravitational analysis, speech, and audio. An important aspect of Time-Frequency (TF) analysis is the identification of separate components in a multi component signal. Highresolution time-frequency (TF) representations and instantaneous frequency (IF)-based methods are needed for analysis, detection and classification of these type signals. TF signal representations enable separation of time-varying components overlapping both in time and frequency domains. It may not be possible to isolate some signal components in one domain using ordinary frequency domain filtering. An important application area of high resolution TF analysis is the radar imaging [1], [2]. The image of the target is obtained from spatial distribution of the reflected signal.

Signal, Image and Video Processing, 2013

Amplitude and phase estimation of AM/FM signals with parametric polynomial representation require... more Amplitude and phase estimation of AM/FM signals with parametric polynomial representation require the polynomial orders for phase and amplitude to be known. But in reality, they are not known and have to be estimated. A well-known method for estimation is the higher-order ambiguity function (HAF) or its variants. But the HAF method has several reported drawbacks such as error propagation and slowly varying or even constant amplitude assumption. Especially for the long duration time-varying signals like AM/FM signals, which require high orders for the phase and amplitude, computational load is very heavy due to nonlinear optimization involving many variables. This paper utilizes a micro-segmentation approach where the length of segment is selected such that the amplitude and instantaneous frequency (IF) is constant over the segment. With this selection first, the amplitude and phase estimates for each micro-segment are obtained optimally in the LS sense, and then, these estimates are concatenated to obtain the overall amplitude and phase estimates. The initial estimates are not optimal but sufficiently close to the optimal solution for subsequent processing. Therefore, by using the initial estimates, the overall polynomial orders for the amplitude and phase are estimated. Using estimated orders, the initial amplitude and phase functions are fitted to the polynomials to obtain the final signal. The method does not use any multivariable nonlinear optimization and is efficient in the sense that the Z. Deprem (B) • A

EURASIP Journal on Advances in Signal Processing, 2011

The problem of component estimation from a multicomponent signal in additive white Gaussian noise... more The problem of component estimation from a multicomponent signal in additive white Gaussian noise is considered. A parametric ML approach, where all components are represented as a multiplication of a polynomial amplitude and polynomial phase term, is used. The formulated optimization problem is solved via nonlinear iterative techniques and the amplitude and phase parameters for all components are reconstructed. The initial amplitude and the phase parameters are obtained via time-frequency techniques. An alternative method, which iterates amplitude and phase parameters separately, is proposed. The proposed method reduces the computational complexity and convergence time significantly. Furthermore, by using the proposed method together with Expectation Maximization (EM) approach, better reconstruction error level is obtained at low SNR. Though the proposed method reduces the computations significantly, it does not guarantee global optimum. As is known, these types of non-linear optimization algorithms converge to local minimum and do not guarantee global optimum. The global optimum is initialization dependent.