• 한국음향학회지
• /
• 제35권4호
• /
• pp.303-309
• /
• 2016

본 논문은 자기부호화기를 이용한 음향신호 분리방법을 제안한다. 사용된 복층구조 신경망 자기부호화기는 입력 신호의 효율적인 표현방법을 자동으로 학습하며, 유사한 특징을 가지고 있는 요소신호들을 군집함으로써 다른 특징의 신호들을 분리할 수 있다. 시간영역과 주파수영역의 변이특성을 추출하기 위하여 단구간푸리에변환(Short-Time Fourier Transform, STFT)을 수행하였으며, 정해진 크기의 사각형 창을 모든 가능한 위치에 적용하여 얻은 단구간 주파수 스펙트럼을 자기부호화기의 입력으로 사용하였다. 자기부호화기의 부호노드들의 값을 이용하여 유사한 스펙트럼 창들을 군집하고, 이를 이용하여 원래의 음원들로 분리해 낼 수 있었다. 분리된 원음들은 원래의 입력신호의 특징을 확실히 나타내었으며, 기존의 비음수 행렬분해(Non-negative Matrix Factorization, NMF) 결과와 주파수 스펙트럼 비교를 통해 그 유효성을 보일 수 있었다.

• Journal of Microbiology and Biotechnology
• /
• 제13권4호
• /
• pp.509-516
• /
• 2003

A model system for the immnunochemical detection of Escherichia coli O157:H7 using a combined immunomagnetic separation (IMS) and test-strip liposome immunoassay (LIA) procedure was developed. Immunomagnetic beads coated with anti-E. coli O157 IgG antibodies were used to separate the E. coli O157 (including the H7 serotype) from culture. Immunoliposomes, whose surface was conjugated to goat anti-E. coli O157:H7 IgG and which encapsulated the marker dye, sulforhodamine B, were used as a detection label. The test strip, onto which antibodies to goat IgG were immobilized, was the immunosensor capturing immunoliposomes that did not bind to E. coli O157:H7 on the immunomagnetic bead-E. coli O157:H7 complexes. In experiments, pure cell culture suspensions of $10^5 E.$ coli O157:H7 organisms per ml produced a measurable signal inhibition, whereas a weak yet detectable signal inhibition occurred with $10^3CFU/ml$. The inhibition signals increased, when the incubation time for IMS was extended to 90 min and higher IgG-tag density (0.4mol%) was used on the liposomes. With 0.2 and 0.4mol% IgG-tagged liposomes, the IMS-LIA procedure showed more improved signal inhibitions than those of a direct (no IMS) LIA. The combined assay, which measures the instantaneous signal from immunoliposomes, can be completed within 90 min, making it significantly faster than conventional plating methods and enzyme-linked immunosorbent assay (ELISA). Accordingly, it is quite feasible to use the combined immunoassay format of IMS and dye-loaded immunoliposomes for the detection of E. coli O157:H7.

• 한국지능시스템학회논문지
• /
• 제14권5호
• /
• pp.623-629
• /
• 2004

In this paper, a Efficient method for the mixed image separation is presented using independent component analysis and the new fast expectation-maximization(EM) algorithm. In general, the independent component analysis (ICA) is one of the widely used statistical signal processing scheme in various applications. However, it has been known that ICA does not establish good performance in source separation by itself. So, Innovation process which is one of the methods that were employed in image separation using ICA, which produces improved the mixed image separation. Unfortunately, the innovation process needs long processing time compared with ICA or EM. Thus, in order to overcome this limitation, we proposed new method which combined ICA with the New fast EM algorithm instead of using the innovation process. Proposed method improves the performance and reduces the total processing time for the Image separation. We compared our proposed method with ICA combined with innovation process. The experimental results show the effectiveness of the proposed method by applying it to image separation problems.

• ETRI Journal
• /
• 제34권6호
• /
• pp.958-961
• /
• 2012

In this letter, we present a new algorithm for the harmonic and percussive separation of jazz music. Using a short-time Fourier transform and nonnegative matrix factorization, the signal is decomposed into rank components. Each component is then split into harmonic and percussive parts using masks calculated based on their tonalities. Finally, the harmonic and percussive parts are separated after applying the masks and a summation. We evaluate the algorithm based on real audio examples using both objective and subjective assessments. The proposed algorithm performs well for the separation of harmonic and percussive parts of jazz excerpts.

• 전기학회논문지
• /
• 제57권8호
• /
• pp.1454-1463
• /
• 2008

Blind source separation by independent component analysis (ICA) has applied in signal processing, telecommunication, and image processing to recover unknown original source signals from mutually independent observation signals. Neural networks are learned to estimate the original signals by unsupervised learning algorithm. Because the outputs of the neural networks which yield original source signals are mutually independent, then mutual information is zero. This is equivalent to minimizing the Kullback-Leibler convergence between probability density function and the corresponding factorial distribution of the output in neural networks. In this paper, we present a learning algorithm using information theory and higher order statistics to solve problem of blind source separation. For computer simulation two deterministic signals and a Gaussian noise are used as original source signals. We also test the proposed algorithm by applying it to several discrete images.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2004년도 학술대회 논문집 정보 및 제어부문
• /
• pp.265-267
• /
• 2004

A blind separation problem in a multiple-input-multiple-output (MIMO) linear time-invariant (LTI) system with finite-alphabet inputs is considered. A discrete-time matrix equation model is used to describe the input-output relation of the system in order to make full use of the advantages of modern digital signal processing techniques. At first, ambiguity problem is investigated. Then, based on the results of the investigation, a new identifiability condition is proposed for the case of an input-data set which is widely used in digital communication. A probability bound such that an arbitrary input matrix satisfies the identifiability condition is derived. Finally, Monte-Carlo simulation is performed to demonstrate the validity of our suggestions.

• 전기학회논문지
• /
• 제56권3호
• /
• pp.629-631
• /
• 2007

In hands-free full-duplex communication systems, acoustic signals picked up by the microphones can be mixed with echo signals as well as noises, which may result in poor performance of the corresponding communication system. Also, the system performance may decrease further if the reverberation occurs since it is harder to estimate the impulse response of the demixing system. For blind source separation (BSS) in such cases, a time-frequency masking approach can be employed to separate undesired echo signals and noises, but, permutation ambiguities also should be solved for the echo cancellation. In this paper, we propose a new acoustic echo cancellation (AEC) approach utilizing the time-frequency masking and higher-order statistics, whereby a desired signal selection, based on coherence and third-order statistics (i.e., kurtosis), is introduced along with output signal normalization. Simulation results demonstrate that the proposed approach yields better echo and noise cancellation performances than the conventional AEC approaches.

• ETRI Journal
• /
• 제28권6호
• /
• pp.761-769
• /
• 2006

In this paper, we address the problem of closely spaced source localization using sensor array processing. In particular, the performance efficiency (measured in terms of the root mean square error) of the unconditional maximum likelihood (UML) algorithm for estimating the direction of arrival (DOA) of near-field sources is evaluated. Four parameters are considered in this evaluation: angular separation among sources, signal-to-noise ratio (SNR), number of snapshots, and number of sources (multiple sources). Simulations are conducted to illustrate the UML performance to compute the DOA of sources in the near-field. Finally, results are also presented that compare the performance of the UML DOA estimator with the existing multiple signal classification approach. The results show the capability of the UML estimator for estimating the DOA when the angular separation is taken into account as a critical parameter. These results are consistent in both low SNR and multiple-source scenarios.

• The Journal of the Acoustical Society of Korea
• /
• 제27권1E호
• /
• pp.30-34
• /
• 2008

An improvement to the existing blind signal separation (BSS) method has been made in this paper. The proposed method models the inherent signal dependency observed in acoustic object to separate the real-world convolutive sound mixtures. The frequency domain approach requires solving the well known permutation problem, and the problem had been successfully solved by a vector representation of the sources whose multidimensional joint densities have a certain amount of dependency expressed by non-spherical distributions. Especially for speech signals, we observe strong dependencies across neighboring frequency bins and the decrease of those dependencies as the bins become far apart. The non-spherical joint density model proposed in this paper reflects this property of real-world speech signals. Experimental results show the improved performances over the spherical joint density representations.

• 한국음향학회지
• /
• 제31권5호
• /
• pp.317-331
• /
• 2012

본 논문에서는 다채널 음향 신호의 음원 분리를 수행하기 위하여, 빔공간-영역에서 다채널 비음수 행렬 분해 기법을 이용하는 음원 분리 시스템을 제안한다. 비음수 행렬 분해(NMF) 기법은 음원 분리에서 최근 널리 쓰이는 알고리즘이며, 특히 최근에는 다채널 비음수 행렬 분해(MC-NMF) 기법으로 발전하여 다채널 음향 신호에 대해서 적용되고 있다. 본 논문에서 제안하는 다채널 비음수 행렬 분해 기법은 빔공간-영역에서 수행되어, 기존의 다채널 비음수 행렬 분해 기법에 비해 좋은 성능을 가진다. 제안되는 비음수 행렬 분해 기법은 SiSEC 2010의 데이터셋을 이용하여 검증되었다.