• The Journal of the Acoustical Society of Korea
• /
• v.22 no.3
• /
• pp.169-174
• /
• 2003

A novel subspace speech enhancement using subband whitening filter is proposed. Previous subspace speech enhancement method either assumes additive white noise or uses whitening filter as a pre-processing for colored noise. The proposed method tries to minimize the signal distortion while reducing residual noise by processing the signal using subband whitening filter. By incorporating the notion of subband whitening filter, spectral resolution in Karhunen-Loeve(KL) domain is improved with the negligible additional computational load. The proposed method outperforms both the subspace method suggested by Ephraim and the spectral subtraction suggested by Boll in terms of segmental signal-to-noise ratio (SNRseg) and perceptual evaluation of speech quality (PESQ).

• Proceedings of the Korea Society for Industrial Systems Conference
• /
• 2006.05a
• /
• pp.111-114
• /
• 2006

비백색잡음(non-white noise)인 잔향(reverberation)이 신호탐지(signal detection)의 주 방해신호인 천해 능동소나(active sonar) 환경에서의 표적탐지는 선백색화기(pre-whitening filter)를 사용하여 수신신호를 백색화한 후 백색잡음에서 최적 탐지기(optimum detector)인 정합필터를 사용한다. 그러나 이 방법은 잔향이 비정상(non-stationary) 특성을 가지기 때문에 구현이 매우 힘들다. 기존의 연구에 따르면 이러한 잔향은 지역적 정상상태(local stationary)라고 가정할 수 있다. 본 논문에서는 먼저 잔향신호의 지역적 정상상태의 범위를 추정(estimation)하고, 이 추정을 바탕으로 천해와 같은 비백색 잔향신호 환경에서 선백색화 블럭 정규화 정합필터(pre-whitening block normalized matched filter)의 성능을 개선할 수 있는 선백색화 기법을 제안하였다. 제안된 잔향신호의 백색화 기법은 표적신호 전 후의 잔향신호를 사용하여 처리블록(processing block)을 백색화하기 때문에 기존의 백색화 기법보다 우수한 성능을 보였다. 제안된 백색화 기법을 이용한 탐지기의 성능을 평가하기 위해 우리나라 인근해역에서 실측된 데이터를 이용하여 컴퓨터 모의실험을 수행하였다. 모의실험 결과 제안된 기법을 사용한 탐지기는 기존의 백색화 기법을 사용한 탐지기보다 우수한 탐지성능을 보였다.

• Journal of Korea Society of Industrial Information Systems
• /
• v.11 no.3
• /
• pp.15-19
• /
• 2006

In shallow water active sonar environment, reverberation which is a non-white noise is one of the main source of performance degradation of target detection. In this case, the received signal is whitened before applying matched filter known as an optimum filter in the presence of white noise. However implementation of this method is very difficult because of the non-stationary characteristic of reverberation. Traditionally reverberation is assumed local stationary. In this paper, we estimate a range of stationary of reverberation signal, and then propose a pre-whitening method which improve the performance of pre-whitening block normalized matched filter in presence of non-white reverberation noise. Proposed whitener shows better whitening performance than traditional whitener because it use later as well as before reverberation of target signal. To evaluate performance of the proposed whitener, an actual measurement data sampled at the East-Sea is used for computer simulation. The target detector with new whitener is shown better performance than detector with traditional whitener.

• Journal of the Institute of Electronics Engineers of Korea TC
• /
• v.46 no.10
• /
• pp.8-13
• /
• 2009

The MF(matched filter) is the optimum signal detector that maximizes the output instantaneous signal power to average noise power ratio in white noise environment. But it cannot give the optimum detection performance if the background noise is not white. So, the whitening process preceding the matched filter is needed in the conventional detector which results in a PWMF(pre-whitening matched filter). Its performance is mainly affected by the estimation accuracy of non-white noise model which is used in the whitening procedure. To estimate more accurate model to improve performance, the computational complexity is increased. Therefore, a spectrum based detector which shows better performance than the PWMF under the similar complexity condition or less complexity under the similar performance condition is proposed in this paper. And its performance and complexity are analyzed and compared with the conventional PWMF.

• Journal of the Institute of Electronics Engineers of Korea TC
• /
• v.45 no.12
• /
• pp.152-158
• /
• 2008

In shallow water environment, the detection performance of an active sonar using matched filter with LFM(linear frequency modulation) pulse can be seriously degraded by reverberation which is considered as non-white noise. To reduce the effect of reverberation, a whitening filter preceding the matched fitter, is usually adopted. In the conventional pre-whitening filter, it is assumed that local stationarity is preserved between detection block and its right ahead block. And then by using the characteristics of the reverberation of preceding block, the reverberation of detection block is estimated and whitened. According to the environment of shallow water, the stationarity of reverberation may be preserved for more blocks. In this case, the reverberation of the detection block can be estimated more accurately if more blocks are used. In this paper, the real reverberation signal which is obtained from shallow sea is analyzed and its proper region of estimation block is examined. And the performance of pre-whitening matched filter is compared and analyzed according to the region of estimation block.

• Proceedings of the Acoustical Society of Korea Conference
• /
• autumn
• /
• pp.201-204
• /
• 2000

음성신호에 섞인 잡음을 처리하기 위해서 단 일 마이크로폰을 이용한 방법이 많이 연구되고 있는데, 그 중에서 Kalman 필터를 이용한 방법은 먼저 음성신호의 모델을 검출하고 잡음이 섞인 신호에서 표준 Kalman 필터를 이용해서 음성신호 성분만을 검출하게 된다. 본 논문에서는 음성신호에 섞인 유색잡음을 백색화하는 방법을 적용하여 Kalman 필터의 잡음제거 성능을 향상시키는 방법을 제안하였다.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2009.05a
• /
• pp.883-887
• /
• 2009

This paper presents a statistical analysis of the MSE of adaptation for the MPDSAP (Maximally polyphase decomposed Subband Affine Projection) algorithm for the an autoregressive (AR) inputs with P order. In subband structure, the Affine Projection (AP) algorithm is transformed to the Normalized Least Mean Square (NLMS) algorithm by applying the polyphase decomposition and the noble identity to the adaptive filter. And also, AR input can be pre-whitened by subband filtering with the Orthonormal Analysis Filters(OAF). In the subband structure, the pre-whitening of the AR(P) inputs provides simple and valid approximations for a statistical analysis of the MSE behaviors for the SAP adaptive filter.

• The Journal of the Acoustical Society of Korea
• /
• v.26 no.3
• /
• pp.129-135
• /
• 2007

In this paper. we propose a prewhitening method for the km reverberation to enhance the target signal. The proposed algorithm uses the dechirping method which inversely compensates the frequency chirp rate of LFM and transforms the LFM reverberation to have stationary frequency property in each data block. Also, using the left and right adjacent beam signals as reference signals. we model frequency response of each data block by AR coefficients. From these coefficients, we implement inverse filter and prewhiten the LFM reverberation of the center beam efficiently.

• Proceedings of the Acoustical Society of Korea Conference
• /
• 1998.06e
• /
• pp.411-414
• /
• 1998

음성의 음질 향상(Speech Enhancement)을 위한 여러 가지 방법 중에서 주파수 차감법(Spectral Subtraction)은 계산량이 적기 때문에 현재 실시간으로 Speech Enhancement를 할 수 있는 가장 적절한 방법이다. 그러나, 이 방법은 원래의 입력음성에 없던 새로운 잡음을 만들어내는 큰 단점이 있는데, 이를 제거하기 위해 많은 연구가 되어오고 있다. 이러한 연구의 방향은 대부분 주변프레임 또는 주변의 주파수 성분과의 평균을 통해 피크값을 무디게 해 줌으로써 새로 생긴 튀는 잡음을 감소시키는 것이다. 이런 방법은 음성자체의 정보 또한 평균이 되어버리게 하는 새로운 단점을 낳는데, 이런 현상은 무성음구간에서 특히 심각해진다. 본 논문에서는 입력음성의 LPC 분석으로 백색필터(Whitening Filter)를 구성하여 이를 통과시킨 잔류신호(Residual)를 주파수 차감하여 얻은 새로운 잔류신호를 역 필터링하여(Synthesis Filter) 개선된 음성을 얻는 방법을 제안하였다. 제안된 알고리듬은, 주파수 차감시 포만트(Formant)의 정보가 더 유지 될 수 있기 때문에 잔류잡음을 줄일 수 있다. 청취 테스트 결과 제안한 방법이 기존의 방법보다 잔류잡음을 더 줄이는 사실을 확인할 수 있었다.

• Korean Journal of Cognitive Science
• /
• v.16 no.4
• /
• pp.323-331
• /
• 2005

This paper proposes an improved system for recognizing facial expressions in various internal states that is illumination-invariant and without detectable rue such as a neutral expression. As a preprocessing to extract the facial expression information, a whitening step was applied. The whitening step indicates that the mean of the images is set to zero and the variances are equalized as unit variances, which reduces murk of the variability due to lightening. After the whitening step, we used the facial expression information based on principal component analysis(PCA) representation excluded the first 1 principle component. Therefore, it is possible to extract the features in the lariat expression images without detectable cue of neutral expression from the experimental results, we ran also implement the various and natural facial expression recognition because we perform the facial expression recognition based on dimension model of internal states on the images selected randomly in the various facial expression images corresponding to 83 internal emotional states.