• 한국통신학회논문지
• /
• 제34권10B호
• /
• pp.1070-1075
• /
• 2009

본 논문에서는 3차원 영상 생성을 위한 깊이맵 추정 및 중간시점 영상합성 방법을 제안한다. 제안하는 방법은 깊이맵의 시간적 상관도를 향상시키기 위해 깊이값을 추정하는 과정에서 기존의 정합 함수에 이전 프레임에서 추정한 깊이값을 고려하는 가중치 함수를 추가한다. 또한, 중간시점 영상을 합성하는 과정에서 발생하는 경계 잡음을 제거하는 방법을 제안한다. 중간시점 영상을 합성할 때, 비폐색 영역을 합성한 후 경계 잡음이 발생할 수 있는 영역을 비폐색 (disocclusion) 영역을 따라 구별한 다음, 잡음이 없는 참조 영상을 이용하여 경계 잡음을 처리한다. 컴퓨터 모의실험 결과를 통해 깊이맵의 시간적 상관도를 향상시켜서 사용자의 시각적 피로감을 줄일 수 있었고, 배경 잡음이 사라진 자연스러운 중간시점 영상을 생성할 수 있었다.

• 한국음향학회지
• /
• 제42권6호
• /
• pp.603-616
• /
• 2023

본 연구는 주파수 및 시간 특성을 활용하여 머신러닝 기반 공동주택 주거소음의 군집화 및 분류를 진행하였다. 먼저, 공동주택 주거소음의 군집화 및 분류를 진행하기 위하여 주거소음원 데이터셋을 구축하였다. 주거소음원 데이터셋은 바닥충격음, 공기전달음, 급배수 및 설비소음, 환경소음, 공사장 소음으로 구성되었다. 각 음원의 주파수 특성은 1/1과 1/3 옥타브 밴드별 L_eq와 L_max값을 도출하였으며, 시간적 특성은 5 s 동안의 6 ms 간격의 음압레벨 분석을 통해 L_eq값을 도출하였다. 공동주택 주거소음원의 군집화는 K-Means clustering을 통해 진행하였다. K-Means의 k의 개수는 실루엣 계수와 엘보우 방법을 통해 결정하였다. 주파수 특성을 통한 주거소음원 군집화는 모든 평가지수에서 3개로 군집되었다. 주파수 특성 기준으로 분류된 각 군집별 시간적 특성을 통한 주거소음원 군집화는 L_eq평가지수의 경우 9개, L_max 경우는 11개로 군집되었다. 주파수 특성을 통해 군집된 각 군집은 타 주파수 대역 대비 저주파 대역의 음에너지의 비율 또한 조사되었다. 이후, 군집화 결과를 활용하기 위한 방안으로 세 종류의 머신러닝 방법을 이용해 주거소음을 분류하였다. 주거소음 분류 결과, 1/3 옥타브 밴드의 L_eq값으로 라벨링된 데이터에서 가장 높은 정확도와 f1-score가 나타났다. 또한, 주파수 및 시간적 특성을 모두 사용하여 인공신경망(Artificial Neural Network, ANN) 모델로 주거소음원을 분류했을 때 93 %의 정확도와 92 %의 f1-score로 가장 높게 나타났다.

• 센서학회지
• /
• 제21권2호
• /
• pp.144-150
• /
• 2012

A microphone for fully implantable hearing device was generally implanted under the skin of the temporal bone. So, the implanted microphone's characteristics can be affected by the accompanying noise due to masticatory movement. In this paper, the implantable microphone with 2-channels structure was designed for reduction of the generated noise signal by masticatory movement. And an experimental model for generation of the noise by masticatory movement was developed with considering the characteristics of human temporal bone and skin. Using the model, the speech signal by a speaker and the artificial noise by a vibrator were supplied simultaneously into the experimental model, the electrical signals were measured at the proposed microphone. The collected signals were processed using a general adaptive filter with least mean square(LMS) algorithm. To confirm performance of the proposed methods, the correlation coefficient and the signal to noise ratio(SNR) before and after the signal processing were calculated. Finally, the results were compared each other.

• 한국소음진동공학회:학술대회논문집
• /
• 한국소음진동공학회 2006년도 추계학술대회논문집
• /
• pp.666-670
• /
• 2006

It has been considered that scattered sounds have a positive effect on a hearing impression of a sound filed. This study investigates the degree and the quality of a scattered sound field by using the acoustical parameters and autocorrelation function(ACF) of impulse responses. The acoustical parameters and fine structure of the ACF of an impulse response were used for the evaluation of the scattered sound field. The relationship between the scattering coefficient of surfaces with various hemisphere diffuser configurations and the acoustical parameters and ACF parameters of impulse responses was investigated.

• 한국방송∙미디어공학회:학술대회논문집
• /
• 한국방송공학회 1995년도 학술대회
• /
• pp.137-141
• /
• 1995

MASF is a kind of temporal filter proposed for noise reduction and temporal band limitation. MASF uses motion vectors to extract temporal information in spatial domain. Therefore, inaccurate motion information causes some distortions in MASF operation. Currently, bilinear interpolation after MBA(Block Matching Algorithm) is used for the motion estimation sheme of MASF. But, this method results in unreliable estimation when the object in image sequence has larger movement than the maximum displacement assumed in BMA or the input images are severely corrupted with noise. In order to solve this problem, we propose a hierarchical motion estimation algorithm for MASF. Experimental results show that the proposed method produces reliable output under large motion and noisy situations.

• 대한음성학회지:말소리
• /
• 제53호
• /
• pp.93-102
• /
• 2005

The performance of a speech recognition system is generally degraded in telephone environment because of distortions caused by background noise and various channel characteristics. In this paper, data-driven temporal filters are investigated to improve the performance of a specific recognition task such as telephone speech. Three different temporal filtering methods are presented with recognition results for Korean connected-digit telephone speech. Filter coefficients are derived from the cepstral domain feature vectors using the principal component analysis. According to experimental results, the proposed temporal filtering method has shown slightly better performance than the previous ones.

• 대한전자공학회논문지SP
• /
• 제45권2호
• /
• pp.74-80
• /
• 2008

시공필터는 공간필터로는 제거할 수 없는 동영상의 노이즈를 제거하지만 알고리듬이 매우 복잡하여 하드웨어로 구현하기에 부적절하다. 본 논문에서는 적응 평균필터 알고리듬을 바탕으로 최대 세 장의 프레임을 사용하는 실시간 시공 노이즈 제거 시스템을 구현한다. 기존의 알고리듬에서 하드웨어로 구현하기에 부적절한 요소들을 수정하였다. 동작 속도를 높이기 위해서 노이즈 추정과 필터링이 병렬적으로 수행되도록 이전 프레임에서 추정된 노이즈를 현재 프레임 필터링에 이용하게 하였다. 또한 필터링 윈도우의 형태를 변형하여 시스템의 동기화를 용이하게 하였다. 제안하는 구조는 Virtex 4 XC4VLX60 상에 구현하였고 총 66%의 슬라이스를 사용하고 최대 80MHz의 속도로 동작하였다.

• Journal of Audiology & Otology
• /
• 제23권2호
• /
• pp.83-88
• /
• 2019

Background and Objectives: Mild cognitive impairment (MCI) is a disorder that usually occurs in the elderly, leading to dementia in some progressive cases. The purpose of this study is to examine the utility of central auditory processing tests as early diagnostic tools for identifying the elderly with MCI. Subjects and Methods: This study was conducted on 20 elderly patients with MCI and 20 healthy matched peers. The speech perception ability in a quiet environment and in the presence of background noise and also temporal resolution were assessed by using Speech Perception in Noise (SPIN) and Gap in Noise (GIN) tests, respectively. Results: The results indicated that the ability to understand speech in a quiet environment did not differ significantly between the two groups. However, SPIN at the three signal-to-noise ratios and the temporal resolution scores were significantly different between the two groups (p<0.001). Conclusions: Individuals with MCI appear to have poorer speech comprehension in noise and a lower temporal resolution than those of the same age, but without cognitive defects. Considering the utility of these tests in identifying cognitive problems, we propose that since the GIN test seems to be less influenced by intervening factors, this test can therefore, be a useful tool for the early screening of elderly people with cognitive problems.

• 대한청각학회지
• /
• 제23권2호
• /
• pp.83-88
• /
• 2019

Background and Objectives: Mild cognitive impairment (MCI) is a disorder that usually occurs in the elderly, leading to dementia in some progressive cases. The purpose of this study is to examine the utility of central auditory processing tests as early diagnostic tools for identifying the elderly with MCI. Subjects and Methods: This study was conducted on 20 elderly patients with MCI and 20 healthy matched peers. The speech perception ability in a quiet environment and in the presence of background noise and also temporal resolution were assessed by using Speech Perception in Noise (SPIN) and Gap in Noise (GIN) tests, respectively. Results: The results indicated that the ability to understand speech in a quiet environment did not differ significantly between the two groups. However, SPIN at the three signal-to-noise ratios and the temporal resolution scores were significantly different between the two groups (p<0.001). Conclusions: Individuals with MCI appear to have poorer speech comprehension in noise and a lower temporal resolution than those of the same age, but without cognitive defects. Considering the utility of these tests in identifying cognitive problems, we propose that since the GIN test seems to be less influenced by intervening factors, this test can therefore, be a useful tool for the early screening of elderly people with cognitive problems.

• 대한원격탐사학회지
• /
• 제37권4호
• /
• pp.719-731
• /
• 2021

This study proposes a two-stage hybrid classification model for crop classification using multi-temporal remote sensing images; the model combines feature embedding by using an autoencoder (AE) with a convolutional neural network (CNN) classifier to fully utilize features including informative temporal and spatial signatures. Long short-term memory (LSTM)-based AE (LAE) is fine-tuned using class label information to extract latent features that contain less noise and useful temporal signatures. The CNN classifier is then applied to effectively account for the spatial characteristics of the extracted latent features. A crop classification experiment with multi-temporal unmanned aerial vehicle images is conducted to illustrate the potential application of the proposed hybrid model. The classification performance of the proposed model is compared with various combinations of conventional deep learning models (CNN, LSTM, and convolutional LSTM) and different inputs (original multi-temporal images and features from stacked AE). From the crop classification experiment, the best classification accuracy was achieved by the proposed model that utilized the latent features by fine-tuned LAE as input for the CNN classifier. The latent features that contain useful temporal signatures and are less noisy could increase the class separability between crops with similar spectral signatures, thereby leading to superior classification accuracy. The experimental results demonstrate the importance of effective feature extraction and the potential of the proposed classification model for crop classification using multi-temporal remote sensing images.