• IEIE Transactions on Smart Processing and Computing
• /
• v.2 no.5
• /
• pp.277-281
• /
• 2013

Modern speaker verification systems based on support vector machines (SVMs) use Gaussian mixture model (GMM) supervectors as their input feature vectors, and the maximum a posteriori (MAP) adaptation is a conventional method for generating speaker-dependent GMMs by adapting a universal background model (UBM). MAP adaptation requires the appropriate amount of input utterance due to the number of model parameters to be estimated. On the other hand, with limited utterances, unreliable MAP adaptation can be performed, which causes adaptation noise even though the Bayesian priors used in the MAP adaptation smooth the movements between the UBM and speaker dependent GMMs. This paper proposes a sparse MAP adaptation method, which is known to perform well in the automatic speech recognition area. By introducing sparse MAP adaptation to the GMM-SVM-based speaker verification system, the adaptation noise can be mitigated effectively. The proposed method utilizes the L0 norm as a regularizer to induce sparsity. The experimental results on the TIMIT database showed that the sparse MAP-based GMM-SVM speaker verification system yields a 42.6% relative reduction in the equal error rate with few additional computations.

• Proceedings of the Korea Multimedia Society Conference
• /
• 2012.05a
• /
• pp.30-33
• /
• 2012

대사성 골 질환인 골다공증(Osteoporosis)의 조기 진단을 위해 X 선 영상에서 골 밀도를 측정하는 방법이 최근 연구되고 있다. 골 밀도는 X 선 영상에서 뼈가 분리되고, 분리된 영역에서의 픽셀에 의해 BMD가 측정되는데, 개선된 영상에서의 정밀한 뼈 추출이 주요한 요소이므로 X 선 영상의 개선은 골다공증의 조기 진단을 위해 필수적이다. 본 논문에서는 sparse 표현법을 도입하여 X 선 영상을 개선시키는 방법을 제안한다. 실험을 통해 제안한 방법의 결과가 기존의 방법인 웨이블릿 BayesShrink에 비해 개선됨을 CNR(Contrast to Noise Ratio)을 통해 확인하였다.

• Proceedings of the KIEE Conference
• /
• 2007.04a
• /
• pp.281-282
• /
• 2007

We address a new representation of DCT/DFT/Wavelet matrices via one hybrid architecture. Based on an element inverse matrix factorization algorithm, we show that the OCT, OFT and Wavelet which based on Haar matrix have the similarrecursive computational pattern, all of them can be decomposed to one orthogonal character matrix and a special sparse matrix. The special sparse matrix belongs to Jacket matrix, whose inverse can be from element-wise inverse or block-wise inverse. Based on this trait, we can develop a hybrid architecture.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2012.04a
• /
• pp.369-372
• /
• 2012

대사성 골 질환인 골다공증(Osteoporosis)의 조기 진단을 위한 골 밀도를 측정하는 방법이 최근 연구되고 있다. 골 밀도 영상은 이중 에너지 X 선 흡수법에 의해 측정되는데, 영상에 존재하는 잡음은 뼈 영역 추출과 골 밀도 계산에 어려움을 주고 있다. 따라서 본 논문에서는 최근 신호처리 분야에서 폭넓게 사용되고 있는 sparse 표현을 도입하여 X 선 영상의 잡음을 제거하는 방법을 제안한다. 실험을 통해 제안한 잡음 제거 방법의 결과가 기존의 방법에 비해 개선됨을 MSR(Mean to Standard deviation Ratio)과 CNR(Contrast to Noise Ratio)을 통해 확인하였다.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.14 no.9
• /
• pp.3745-3761
• /
• 2020

Low-rank representation methods already achieve many applications in the image reconstruction. However, for high-gradient image patches with rich texture details and strong edge information, it is difficult to find sufficient similar patches. Existing low-rank representation methods usually destroy image critical details and fail to preserve edge structure. In order to promote the performance, a new representation-based image super-resolution reconstruction method is proposed, which combines gradient domain guided image filter with the structure-constrained low-rank representation so as to enhance image details as well as reveal the intrinsic structure of an input image. Firstly, we extract the gradient domain guided filter of each atom in high resolution dictionary in order to acquire high-frequency prior information. Secondly, this prior information is taken as a structure constraint and introduced into the low-rank representation framework to develop a new model so as to maintain the edges of reconstructed image. Thirdly, the approximate optimal solution of the model is solved through alternating direction method of multipliers. After that, experiments are performed and results show that the proposed algorithm has higher performances than conventional state-of-the-art algorithms in both quantitative and qualitative aspects.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.8 no.6
• /
• pp.1946-1963
• /
• 2014

Most background subtraction methods focus on dynamic and complex scenes without considering robustness against noise. This paper proposes a background subtraction algorithm based on dictionary learning and sparse coding for handling low light conditions. The proposed method formulates background modeling as the linear and sparse combination of atoms in the dictionary. The background subtraction is considered as the difference between sparse representations of the current frame and the background model. Assuming that the projection of the noise over the dictionary is irregular and random guarantees the adaptability of the approach in large noisy scenes. Experimental results divided in simulated large noise and realistic low light conditions show the promising robustness of the proposed approach compared with other competing methods.

• Structural Engineering and Mechanics
• /
• v.85 no.1
• /
• pp.119-133
• /
• 2023

Impact event is the key factor influencing the operational state of the mechanical equipment. Additionally, nonlinear factors existing in the complex mechanical equipment which are currently attracting more and more attention. Therefore, this paper proposes a novel hybrid-separate identification strategy to solve the force identification problem of the nonlinear structure under impact excitation. The 'hybrid' means that the identification strategy contains both l1-norm (sparse) and l2-norm regularization methods. The 'separate' means that the nonlinear response part only generated by nonlinear force needs to be separated from measured response. First, the state-of-the-art two-step iterative shrinkage/thresholding (TwIST) algorithm and sparse representation with the cubic B-spline function are developed to solve established normalized sparse regularization model to identify the accurate impact force and accurate peak value of the nonlinear force. Then, the identified impact force is substituted into the nonlinear response separation equation to obtain the nonlinear response part. Finally, a reduced transfer equation is established and solved by the classical Tikhonove regularization method to obtain the wave profile (variation trend) of the nonlinear force. Numerical and experimental identification results demonstrate that the novel hybrid-separate strategy can accurately and efficiently obtain the nonlinear force and impact force for the nonlinear structure.

• Journal of the Chungcheong Mathematical Society
• /
• v.15 no.2
• /
• pp.97-107
• /
• 2003

An efficient algorithm for the multiplication in a binary finite filed using a normal basis representation of $F_{2^m}$ is discussed and proposed for software implementation of elliptic curve cryptography. The algorithm is developed by using the storage scheme of sparse matrices.

• Journal of IKEEE
• /
• v.19 no.3
• /
• pp.342-348
• /
• 2015

High speed implementation of digital filters is required in high data rate applications such as hard-wired wide band modem and high resolution video codec. Since the critical path of the digital filter is the MAC (multiplication and accumulation) circuit, the filter coefficient with sparse non-zero bits enables high speed implementation with adders of low hardware cost. Compressive sensing has been reported to be very successful in sparse representation and sparse signal recovery. In this paper a filter design method for digital FIR filters with CSD (canonic signed digit) coefficients using compressive sensing technique is proposed. The sparse non-zero signed bits are selected in the greedy fashion while pruning the mistakenly selected digits. A few design examples show that the proposed method can be utilized for designing sparse CSD coefficient digital FIR filters approximating the desired frequency response.

• Smart Structures and Systems
• /
• v.25 no.3
• /
• pp.369-384
• /
• 2020

A wavefield sparse reconstruction technique based on compressed sensing is developed in this work to dramatically reduce the number of measurements. Firstly, a severely underdetermined representation of guided wavefield at a snapshot is established in the spatial domain. Secondly, an optimal compressed sensing model of guided wavefield sparse reconstruction is established based on l₁-norm penalty, where a suite of discrete cosine functions is selected as the dictionary to promote the sparsity. The regular, random and jittered undersampling schemes are compared and selected as the undersampling matrix of compressed sensing. Thirdly, a gradient projection method is employed to solve the compressed sensing model of wavefield sparse reconstruction from highly incomplete measurements. Finally, experiments with different excitation frequencies are conducted on an aluminum plate to verify the effectiveness of the proposed sparse reconstruction method, where a scanning laser Doppler vibrometer as the true benchmark is used to measure the original wavefield in a given inspection region. Experiments demonstrate that the missing wavefield data can be accurately reconstructed from less than 12% of the original measurements; The reconstruction accuracy of the jittered undersampling scheme is slightly higher than that of the random undersampling scheme in high probability, but the regular undersampling scheme fails to reconstruct the wavefield image; A quantified mapping relationship between the sparsity ratio and the recovery error over a special interval is established with respect to statistical modeling and analysis.