• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.7 no.4
• /
• pp.649-655
• /
• 2006

This paper proposes a HMM-based recognition method using DMSVQ(Dynamic Multi-Section Vector Quantization) codebook by DMS model and double spectral feature, as a method on the speech recognition of speaker-independent. LPC cepstrum parameter is used as a instantaneous spectral feature and LPC cepstrum's regression coefficient is used as a dynamic spectral feature These two spectral features are quantized as each VQ codebook. HMM using DMS model is modeled by receiving instantaneous spectral feature and dynamic spectral feature by input. Other experiments to compare with the results of recognition experiments using proposed method are implemented by the various conventional recognition methods under the equivalent environment of data and conditions. Through the experiment results, it is proved that the proposed method in this paper is superior to the conventional recognition methods.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2003.11a
• /
• pp.493-496
• /
• 2003

본 논문에서는 계산의 복잡성을 단순화하고, 얼굴영상에 대해 높은 얼굴 인식률을 얻기 위해 2D-HMM(Midden Markov Model) 얼굴인식 방법을 제안하고 실험하였다. 계산의 복잡성을 줄이기 위해 기존의 픽셀값 대신에 2D-DCT계수를 관측벡터로 사용함으로써 관측벡터의 크기와 인식 시스템의 복잡성을 줄일 수 있었다. 얼굴인식 시스템의 성능을 평가하기 위하여 Yale, ORL의 얼굴 데이터베이스에 대하여 기존의 얼굴인식 방법으로 널리 알려진 Eigenface 방법, LDA 방법과 본 논문에서 제안한 방법인 1D-HMM, 2D-HMM방법의 인식률을 비교 평가하였다. 실험결과 2D-HMM 방법의 인식률이 99.5%로 기존의 얼굴인식 방법들보다 우수한 성능을 나타냈다. 또한 일정 state수에 대해 mixture의 수가 증가할수록 인식결과가 좋아짐을 알 수 있었다.

• Annual Conference on Human and Language Technology
• /
• 2005.10a
• /
• pp.7-12
• /
• 2005

본 논문은 은닉 마코프 모델(hidden Markov Model: HMM)을 2 단계로 적용하여 논문 모집공고(Call-for-Paper: CFP)에서 필요한 정보를 추출하는 방법을 제안한다. HMM은 순차적인 흐름의 정보를 담고 있는 데이터를 잘 설명할 수 있으며 CFP가 담고 있는 정보에는 순서가 있기 때문에, CFP를 HMM으로 설명할 수 있다. 하지만, 문서를 전체적으로(global) 파악하는 HMM만으로는 정보의 정확한 경계를 파악할 수 없다. 따라서 첫 번째 단계로 CFP문서에서 구(phrase) 단위를 구성하는 단어의 열에 대한 HMMs을 통해 국부적으로(local) 정보의 경계와 대강의 종류를 파악한다. 그리고 두 번째 단계에서 전체적인 문서의 내용 흐름에 근거하여 구축된 HMM을 이용하여 그 정보가 세부적으로 어떤 종류의 정보인지 정한다. PASCAL challenge에서 제공받은 Cff 말뭉치에 대한 첫 번째 단계의 실험 결과, 0.60의 재현률과 0.61의 정확률을 보였으며, 정확률과 재현률을 바탕으로 F-measure를 측정한 결과 0.60이었다.

• The Journal of the Acoustical Society of Korea
• /
• v.21 no.7
• /
• pp.641-647
• /
• 2002

In this paper, we propose the reduction method of the number of parameters in the segmental-feature HMM using trend quantization method. The proposed method shares the trend information of the polynomial trajectories by quantization. The trajectory is obtained by the sequence of feature vectors of speech signals and can be divided by trend and location information. The trend indicates the variation of consequent frame features, while the location points to the positional difference of the trajectories. Since the trend occupies the large portion of SFHMM, if the trend is shared, the number of parameters maybe decreases. To exploit the proposed system the experiments are performed on TIMIT corpus. The experimental results show that the performance of the proposed system is roughly similar to that of previous system. Therefore, the proposed system can be considered one of parameter reduction method.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.13 no.6
• /
• pp.673-678
• /
• 2003

In this paper, we proposes a method for face recognition using HMM(hidden Markov model) and wavelet coefficients First, input images are compressed by using the multi-resolution analysis based on the discrete wavelet transform. And then, the wavelet coefficients obtained from each subband are used as feature vectors to construct the HMMs. In the recognition stage, we obtained higher recognition rate by summing of each recognition rate of wavelet subband. The usefulness of the proposed method was shown by comparing with conventional VQ and DCT-HMM ones. The experimental results show that the proposed method is more satisfactory than previous ones.

• Journal of Electrical Engineering and Technology
• /
• v.7 no.1
• /
• pp.109-114
• /
• 2012

In this paper, we utilize training strategy of hidden Markov model (HMM) to use in versatile issues such as classification of time-series sequential data such as electric transient disturbance problem in power system. For this, an automatic means of optimizing HMMs would be highly desirable, but it raises important issues: model interpretation and complexity control. With this in mind, we explore the possibility of using genetic algorithm (GA) and harmony search (HS) algorithm for optimizing the HMM. GA is flexible to allow incorporating other methods, such as Baum-Welch, within their cycle. Furthermore, operators that alter the structure of HMMs can be designed to simple structures. HS algorithm with parameter-setting free technique is proper for optimizing the parameters of HMM. HS algorithm is flexible so as to allow the elimination of requiring tedious parameter assigning efforts. In this paper, a sequential data analysis simulation is illustrated, and the optimized-HMMs are evaluated. The optimized HMM was capable of classifying a sequential data set for testing compared with the normal HMM.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2002.04a
• /
• pp.647-650
• /
• 2002

The human faces do not have distinct features unlike other general objects. In general the features of eyes, nose and mouth which are first recognized when human being see the face are defined. These features have different characteristics depending on different human face. In this paper, We propose a face recognition algorithm using the hidden Markov model(HMM). In the preprocessing stage, we find edges of a face using the locally adaptive threshold scheme and extract features based on generic knowledge of a face, then construct a database with extracted features. In training stage, we generate HMM parameters for each person by using the forward-backward algorithm. In the recognition stage, we apply probability values calculated by the HMM to input data. Then the input face is recognized by the euclidean distance of face feature vector and the cross-correlation between the input image and the database image. Computer simulation shows that the proposed HMM algorithm gives higher recognition rate compared with conventional face recognition algorithms.

• Proceedings of the KIEE Conference
• /
• 2007.07a
• /
• pp.1860-1861
• /
• 2007

주식 예측은 주식 시장이 생긴 이래로 투자자들이나, 금융 전문가들 사이에서 매우 중요한 일이 되어 왔다. 그러한 중요성으로 인해 엘리오트 파동이론과 같은 많은 주식 예측 기법이 제시되었고, 또한 이러한 예�G의 자동화를 위해 인공지능분야에서도 많은 연구가 있어왔다. 주가 예측에 패턴인식 방법을 적용한 기존의 연구로는 주로 ANN(Artificial Neural Network)방식과 은닉 마코프 모델(HMM, Hidden Markov Model)이 있었고, 본 논문에서는 HMM을 이용한 방법을 제안한다. HMM은 시간 순차적인 패턴을 가지는 모델의 인식에 좋은 성능을 보여 주로 음성인식 분야에서 많이 이용되고 있다. 주식 변화 역시 시간 순차적 흐름에 따라 기울기의 변화가 어느 정도 일정한 패턴을 가지는 성질이 있고, 이것은 HMM을 이용한 패턴인식으로 주식의 앞으로의 변화를 예측하기에 적합한 요인이 된다. 본 논문에서는 이를 위해 다음과 같은 과정을 걸쳤다. 첫 번째로 실존 회사의 장기간의 주식 테이터를 기반으로 여러 개의 HMM모델을 학습 하였다. 두 번째로 예측하고자 하는 기간 이전의 주식 변화 데이터를 입력으로 하여, 이전에 이와 유사한 패턴이 있었는지를 HMM을 통해 알아냈다. 마지막으로 이렇게 알아낸 패턴을 이용하여 앞으로의 주식 변화를 예측하였다. 실험은 실제 주식 변화와 예측값의 비교를 통해 정확도를 검증하였다.

• The Journal of the Acoustical Society of Korea
• /
• v.25 no.4
• /
• pp.159-163
• /
• 2006

In this paper a data-driven method to improve the performance of the Jacobian adaptation (JA) for the noisy speech recognition is proposed. In stead of constructing the reference HMM by using the model composition method like the parallel model combination (PMC), we propose to train the reference HMM directly with the noisy speech. This was motivated from the idea that the directly trained reference HMM will model the acoustical variations due to the noise better than the composite HMM. For the estimation of the Jacobian matrices, the Baum-Welch algorithm is employed during the training. The recognition experiments have been done to show the improved performance of the proposed method over the Jacobian adaptation as well as other model compensation methods.

• Journal of the Korean Institute of Telematics and Electronics C
• /
• v.36C no.8
• /
• pp.46-53
• /
• 1999

We have developed glove based hand gesture recognition system for recognizing 3D gesture of operators in remote work environment. Polhemus sensor attached to the PinchGlove is employed to obtain the sequence of 3D positions of a hand trajectory. These 3D data are then encoded as the input to our recognition system. We propose the use of the strokes to be modeled by HMMs as basic units. The gesture models are constructed by concatenating stroke HMMs and thereby the HMMs for the newly defined gestures can be created without retraining their parameters. Thus, by using stroke models rather than gesture models, we can raise the system extensibility. The experiment results for 16 different gestures show that our stroke based composite HMM performs better than the conventional gesture based HMM.