신경망 기반의 신호 천이 고속 분류 연구

장연수 한국정보기술학회 2022 한국정보기술학회논문지 Vol.20 No.6

In cognitive radio communication systems, classification ability is required for an individual communication device for adaptive reuse of spectrum. If characteristics of a transient signal that is the initial duration of a transmitted communication signal can be analyzed, we can classify the individual transmitter. For a real-time operation, a fast classification with low complexity should be available. We propose a fast classification method of a communication signal transient based on a neural network. To improve the processing speed of the classifier with low complexity, we use transient samples of the communication signal without preprocessing and adopt a simple neural network structure. Classification is performed for twelve transient types including raised-cosine and square-root raised-cosine shapes. Through computer simulation, it was confirmed that the proposed method showed a classification accuracy of 99% at a signal-to-noise ratio of 10 dB. Furthermore, because the proposed method has a structure in which multiplication is performed 12 times per clock, it was analyzed as having low computational complexity. 인지 무선 통신 시스템에서는 가변적인 스펙트럼 재사용을 위해 개별 통신 객체를 분류하는 능력이 요구된다. 송신되는 통신 신호의 초기 구간인 천이 신호의 특성을 분석하면 개별 송신기를 분류할 수 있으며, 실시간으로 운용되기 위해서는 낮은 계산 복잡도를 갖고 고속의 분류가 가능해야 한다. 본 논문에서는 신경망 기반의 통신 신호 천이 고속 분류 방법을 제안한다. 분류기의 복잡도를 낮추면서 처리 속도를 향상시키기 위해 전처리 없이 통신 신호 천이 구간의 샘플을 사용하고 단순한 형태의 신경망을 적용한다. 상승 코사인 및 제곱근 상승 코사인 천이 타입을 포함하여 12가지 종류의 천이 타입에 대한 분류를 수행한다. 컴퓨터 모의실험을 통해 제안하는 방법이 신호 대 잡음비 10 dB에서 99%의 분류 정확도를 나타내는 것을 확인하였다. 그리고 제안하는 방법은 매 클럭당 12번의 곱셈을 수행하는 구조이므로, 낮은 계산 복잡도를 갖는 것으로 분석되었다.

영상 융합 기술을 이용한 색 번짐 개선 방법

장준영(Joonyoung Chang),강문기(Moon Gi Kang) 대한전자공학회 2008 電子工學會論文誌-SP (Signal processing) Vol.45 No.4

본 논문에서는 대역 제한된 색도 신호에 의해 발생하는 색 번짐 현상을 TV 수신단이나 MPEG 디코더에서 효과적으로 개선하는 색 번짐 개선 방법을 제안한다. 비디오 영상 신호는 영상의 밝기 정보를 가지고 있는 한 개의 휘도 신호와 색상 정보를 가지고 있는 두 개의 색도 신호로 구성되어 있으며 사람의 눈이 미세한 면적에 대해서는 색의 변화를 거의 인식하지 못하는 점을 이용하여 색도 신호의 고주파 정보를 제한하고 있다. 하지만 HDTV와 같은 고화질 화상 제품이 생산됨에 따라 색 번짐 현상이 화질 저하의 요인으로 인식되기 시작하였다. 본 논문에서는 영상 융합 기술을 이용하여 색도 신호보다 더 많은 고주파 정보를 가지고 있는 휘도 신호의 고주파 성분을 이용하여 손상된 색도 신호의 고주파 성분을 추정하는 방법을 제안한다. 휘도 신호로부터 추출된 고주파 성분은 추정된 색도 신호와 원본 색도 신호 사이에 발생하는 에러의 l₂-norm이 최소화 되도록 설계된 공간 적응적 가중치에 의해서 적절히 조절된 후에 입력 색도 신호와 결합되어 색 번짐이 개선된 색도 신호를 얻을 수 있게 한다. 제안하는 색 번짐 개선 방법은 색도 신호의 해상도를 휘도 신호의 해상도만큼 향상시키기 때문에 자연스러운 결과를 출력하며 또한 기존의 방법에서는 개선하기 힘들었던 좁은 경계에서의 색 번짐 현상도 효과적으로 개선한다. 실험 결과에서는 제안된 방법이 기존의 방법에 비해 시각적 및 수치적인 면에서 뛰어난 결과를 보임을 확인할 수 있다. In this paper, we propose a color transient improvement (CTI) algorithm based on image fusion to improve the color transient in the television(TV) receiver or in the MPEG decoder. Video image signals are composed of one luminance and two chrominance components, and the chrominance signals have been more band-limited than the luminance signals since the human eyes usually cannot perceive changes in chrominance over small areas. However, nowadays, as the advanced media like high-definition TV(HDTV) is developed, the blurring of color is perceived visually and affects the image quality. The proposed CTI method improves the transient of chrominance signals by exploiting the high-frequency information of the luminance signal. The high-frequency component extracted from the luminance signal is modified by spatially adaptive weights and added to the input chrominance signals. The spatially adaptive weight is estimated to minimize the l₂-norm of the error between the original and the estimated chrominance signals in a local window. Experimental results with various test images show that the proposed algorithm produces steep and natural color edge transition and the proposed method outperforms conventional algorithms in terms of both visual and numerical criteria.

베이즈 분류기를 이용한 수중 배경소음하의 과도신호 분류

김주호(Juho Kim),복태훈(Tae-Hoon Bok),팽동국(Dong-Guk Paeng),배진호(Jinho Bae),이종현(Chong Hyun Lee),김성일(Seongil Kim) 한국해양공학회 2012 韓國海洋工學會誌 Vol.26 No.4

In this paper, a Bayesian classifier based on PCA (principle component analysis) is proposed to classify underwater transient signals using 16<SUP>th</SUP> order LPC<linear predictive coding> coefficients as feature vector. The proposed classifier is composed of two steps. The mechanical signals were separated from biological signals in the first step, and then each type of the mechanical signal was recognized in the second step. Three biological transient signals and two mechanical signals were used to condut esperiments. The classification ratios for the feature vectors of biological signals and mechanical signals were 94.75% and 97.23, respectively, when all 16 order LPC vector were used. In order to determine the effect of underwater noise on the classification performance, underwater ambient noise was added to the test signals and the classification ratio according to SNR (signal-to-noise ratio) was compared by changing dimension of feature vector using PCA. The classification ratios of the biological and mechanical signals under ocean ambient noise at 10dB SNR, were 0.51% and 100% respectively. However, the ratios were changed to 53.07%and 83.14% when the dimension of feature vector was converted to three by applying PCA. For correct, classification, it is required SNR over 10dB for three dimension feature vector and over 30dB SNR for seven dimension feature vector under ocean ambient noise environment.

MFCC 특징 벡터를 이용한 수중 천이 신호 식별

임태균,황찬식,이형욱,배건성,Lim, Tae-Gyun,Hwang, Chan-Sik,Lee, Hyeong-Uk,Bae, Keun-Sung 한국통신학회 2007 韓國通信學會論文誌 Vol.32 No.8c

일반적으로 천이 신호의 식별은 지진학이나 상태 모니터링 분야, 특히 수중 음향 신호 처리 분야에서 활발한 연구가 이루어지고 있다. 수중 환경에서 발생하는 천이 신호로는 돌고래와 같은 해양 생물이 내는 천이 신호와 선박, 잠수함 등에서 발생하는 인위적인 천이 신호 등이 있으며, 수중 감시 체계에서 이러한 수중 천이 신호를 식별하는 문제는 매우 중요한 연구 주제이다. 본 논문에서는 음성 인식 분야에서 우수한 인식 성능을 보이는 MFCC(Mel Frequency Cepstral Coefficient)를 기반으로, 천이 신호로 탐지된 입력 신호에 대하여 분석 프레임 단위로 MFCC 특징 벡터를 추출하고, 식별하고자 하는 데이터베이스에 있는 모든 참조 신호들의 MFCC 특징 벡터와의 유클리디언 거리(euclidean distance)를 계산한 후, 가장 작은 값을 갖는 참조 신호로 입력 프레임들을 사상(mapping)시킴으로써 사상이 가장 많이 된 참조 신호로 탐지된 수중 천이신호를 식별하는 프레임 기반의 식별 알고리즘을 제안한다. This paper presents a new method for classification of underwater transient signals, which employs frame-based decision with Mel Frequency Cepstral Coefficients(MFCC). The MFCC feature vector is extracted frame-by-frame basis for an input signal that is detected as a transient signal, and Euclidean distances are calculated between this and all MFCC feature. vectors in the reference database. Then each frame of the detected input signal is mapped to the class having minimum Euclidean distance in the reference database. Finally the input signal is classified as the class that has maximum mapping rate in the reference database. Experimental results demonstrate that the proposed method is very promising for classification of underwater transient signals.

Pilot Study on Prediction of Human Hand Configuration Using Transient State of Surface-Electromyography Signals

MinKyu Kim,Keehoon Kim 제어로봇시스템학회 2013 제어로봇시스템학회 국제학술대회 논문집 Vol.2013 No.10

Surface electromyography (sEMG) signals have been applied as control commands in numerous humanrobot interface systems and have been deployed for rehabilitation or clinical applications. Although lots of previous workers have tried to determine features appropriate for specific sEMG-signal classification problems, little of this work has involved deeply searching for the inner characteristics of the signals. In this study, we try to evaluate the properties of the transient state of sEMG signals on randomly mounted, dry-type electrodes and use this to rapidly predict three kinds of hand configurations - rock, scissors and paper motions. In experiments, subjects performed a rock-scissor-paper game with a virtual hand. For data acquisition, the sEMG signals were sampled at 1 kHz with eightchannel electrodes (wearable, dry type) that were randomly mounted on forearms [2]. The results verified that the proposed algorithm, using the property of the transient state of sEMG signals, works successfully.

Underwater Transient Signal Classification Using Binary Pattern Image of MFCC and Neural Network

LIM, Taegyun,BAE, Keunsung,HWANG, Chansik,LEE, Hyeonguk The Institute of Electronics, Information and Comm 2008 IEICE transactions on fundamentals of electronics, Vol.91 No.3

<P>This paper presents a new method for classification of underwater transient signals, which employs a binary image pattern of the mel-frequency cepstral coefficients as a feature vector and a feed-forward neural network as a classifier. The feature vector is obtained by taking DCT and 1-bit quantization for the square matrix of the mel-frequency cepstral coefficients that is derived from the frame based cepstral analysis. The classifier is a feed-forward neural network having one hidden layer and one output layer, and a back propagation algorithm is used to update the weighting vector of each layer. Experimental results with underwater transient signals demonstrate that the proposed method is very promising for classification of underwater transient signals.</P>

프레임 기반의 효율적인 수중 천이신호 식별을 위한 참조 신호의 벡터 양자화

임태균(Tae Gyun Lim),김태환(Tae Hwan Kim),배건성(Keun Sung Bae),황찬식(Chan Sik Hwang) 한국통신학회 2009 韓國通信學會論文誌 Vol.34 No.2C

프레임 단위로 식별 데이터베이스에 저장된 참조 신호의 특징 벡터와 유사성을 비교하여 입력 신호를 식별하는 경우에, 참조 신호의 특징 벡터로 데이터베이스를 어떻게 구성하는가에 따라 식별 성능은 영향을 받을 수 있다. 즉, 식별 데이터베이스의 구성 방법에 따라 데이터베이스의 크기와 식별을 위한 계산량, 식별 성능 등이 결정되며, 이것은 실제로 수중 천이신호 식별 시스템을 구성할 때 중요한 문제이다. 본 논문에서는 LBG 벡터 양자화 기법을 이용하여 식별 데이터베이스의 크기를 줄여줌으로써 프레임 기반 수중 천이신호 식별 기법의 효율성을 증가시킬 수 있는 방법을 제안하고, 실험을 통하여 제안한 방법의 타당성을 검증하였다. When we classify underwater transient signals with frame-by-frame decision, a database design method for reference feature vectors influences on the system performance such as size of database, computational burden and recognition rate. In this paper the LBG vector quantization algorithm is applied to reduction of the number of feature vectors for each reference signal for efficient classification of underwater transient signals. Experimental results have shown that drastic reduction of the database size can be achieved while maintaining the classification performance by using the LBG vector quantization.

Blockade of Apoptosis Signal-Regulating Kinase 1 Attenuates Matrix Metalloproteinase 9 Activity in Brain Endothelial Cells and the Subsequent Apoptosis in Neurons after Ischemic Injury

Cheon, So Y.,Cho, Kyoung J.,Kim, So Y.,Kam, Eun H.,Lee, Jong E.,Koo, Bon-Nyeo Frontiers Media S.A. 2016 Frontiers in cellular neuroscience Vol.10 No.-

<P>Conditions of increased oxidative stress including cerebral ischemia can lead to blood–brain barrier dysfunction via matrix metalloproteinase (MMP). It is known that MMP-9 in particular is released from brain endothelial cells is involved in the neuronal cell death that occurs after cerebral ischemia. In the intracellular signaling network, apoptosis signal-regulating kinase 1 (ASK1) is the main activator of the oxidative stress that is part of the pathogenesis of cerebral ischemia. ASK1 also promotes apoptotic cell death and brain infarction after ischemia and is associated with vascular permeability and the formation of brain edema. However, the relationship between ASK1 and MMP-9 after cerebral ischemia remains unknown. Therefore, the aim of the present study was to determine whether blocking ASK1 would affect MMP-9 activity in the ischemic brain and cultured brain endothelial cells. Our results showed that ASK1 inhibition efficiently reduced MMP-9 activity <I>in vivo</I> and <I>in vitro</I>. In endothelial cell cultures, ASK1 inhibition upregulated phosphatidylinositol 3-kinase/Akt/nuclear factor erythroid 2 [NF-E2]-related factor 2/heme oxygenase-1 signals and downregulated cyclooxygenase-2 signals after hypoxia/reperfusion. Additionally, in neuronal cell cultures, cell death occurred when neurons were incubated with endothelial cell-conditioned medium (EC-CM) obtained from the hypoxia/reperfusion group. However, after incubation with EC-CM and following treatment with the ASK1 inhibitor NQDI-1, neuronal cell death was efficiently decreased. We conclude that suppressing ASK1 decreases MMP-9 activity in brain endothelial cells, and leads to decreased neuronal cell death after ischemic injury.</P>

웨이브렛 패킷 기반 캡스트럼 계수를 이용한 수중 천이신호 특징 추출 알고리즘

김주호(Juho Kim),팽동국(Dong-Guk Paeng),이종현(Chong Hyun Lee),이승우(Seung Woo Lee) 한국해양공학회 2014 韓國海洋工學會誌 Vol.28 No.6

In general, the number of underwater transient signals is very limited for research on automatic recognition. Data-dependent feature extraction is one of the most effective methods in this case. Therefore, we suggest WPCC (Wavelet packet ceptsral coefficient) as a feature extraction method. A wavelet packet best tree for each data set is formed using an entropy-based cost function. Then, every terminal node of the best trees is counted to build a common wavelet best tree. It corresponds to flexible and non-uniform filter bank reflecting characteristics for the data set. A GMM (Gaussian mixture model) is used to classify five classes of underwater transient data sets. The error rate of the WPCC is compared using MFCC (MeI-frequency ceptsral coefficients). The error rates of WPCC-db20, db40, and MFCC are 0.4%, 0%, and 0.4%, respectively, when the training data consist of six out of the nine pieces of data in each class. However, WPCC-db20 and db40 show rates of 2.98% and 1.20%, respectively, while MFCC shows a rate of 7.14% when the training data consists of only three pieces. This shows that WPCC is less sensitive to the number of training data pieces than MFCC. Thus, it could be a more appropriate method for underwater transient recognition. These results may be helpful to develop an automatic recognition system for an underwater transient signal.

Examining the Function of Wavelet Packet Transform (WPT) and Continues Wavelet Transform (CWT) in Recognizing the Crack Specification

Mohammad Ali Lotfollahi-Yaghin,Mahdi Koohdaragh 대한토목학회 2011 KSCE JOURNAL OF CIVIL ENGINEERING Vol.15 No.3

Modern and efficient methods focus on signal analysis and have drawn researchers' attention to it in recent years. These methods mainly include Continuous Wavelet and Wavelet Packet transforms. The main advantage of the application of these Wavelets is their ability to analyze the signal position in different times and places. The frequency decomposition of this transform in location with high frequencies is very poor. Wavelet packet transform is more advanced form of continuous wavelet and can make a perfect level by level resolution for each signal. However, very few studies have been done in this field. In the present study, first there was an attempt to do a modal analysis on the structure by the ANSYS finite elements software, then using MATLAB, the wavelet was investigated through a continuous wavelet analysis. Finally, the results were displayed in 2-D location-coefficient figures. In the second form, transient dynamic analysis was done on the structure and to find out the characteristics of the crack, wavelet packet energy rate index was suggested. The results revealed that suggested index in the second form is both practical and applicable.