• ETRI Journal
• /
• 제46권1호
• /
• pp.22-34
• /
• 2024

Exposure to varied noisy environments impairs the recognition performance of artificial intelligence-based speech recognition technologies. Degraded-performance services can be utilized as limited systems that assure good performance in certain environments, but impair the general quality of speech recognition services. This study introduces an audiovisual speech recognition (AVSR) model robust to various noise settings, mimicking human dialogue recognition elements. The model converts word embeddings and log-Mel spectrograms into feature vectors for audio recognition. A dense spatial-temporal convolutional neural network model extracts features from log-Mel spectrograms, transformed for visual-based recognition. This approach exhibits improved aural and visual recognition capabilities. We assess the signal-to-noise ratio in nine synthesized noise environments, with the proposed model exhibiting lower average error rates. The error rate for the AVSR model using a three-feature multi-fusion method is 1.711%, compared to the general 3.939% rate. This model is applicable in noise-affected environments owing to its enhanced stability and recognition rate.

• 약학회지
• /
• 제46권5호
• /
• pp.331-336
• /
• 2002

The purpose of this study is to develop a non-invasive blood glucose measurement method by a portable near infrared (NIR) system which was newly integrated by our lab. The portable NIR system includes a tungsten halogen lamp, a specialized reflectance fiber optic probe and a photo diode array type InGaAs detector; which was developed by a microchip technology based on the lithography. Reflectance NIR spectra of different parts of human body (finger tip, earlobe, and inner lip) were recorded by using a fiber optic probe. The spectra were collected over the spectral range 1100 ∼ 1740 nm. Partial least squares regression (PLSR) was applied for the calibration and validation for the determination of blood glucose. The calibration model from earlobe spectra presented better results, showing good correlation with a glucose oxidase method which is a mostly used standard method. This model predicted the glucose concentration for validation set with a SEP of 33 mg/dL. This study indicated the feasibility for non-invasive monitoring of blood glucose by a portable near infrared system.

• 한국정보과학회논문지:소프트웨어및응용
• /
• 제27권6호
• /
• pp.618-631
• /
• 2000

본 논문에서는 2차원 얼굴 영상의 움직임을 추출하여 3차원 얼굴 모델에 합성하는 방법을 연구하였다. 본 논문은 동영상에서의 움직임을 추정하기 위하여 광류를 기반으로 한 추정방법을 이용하였다. 2차원 동영상에서 얼굴요소 및 얼굴의 움직임을 추정하기 위해 인접한 두 영상으로부터 계산된 광류를 가장 잘 고려하는 매개변수화된 움직임 벡터들을 추출한다. 그리고 나서, 이를 소수의 매개변수들의 조합으로 만들어 얼굴의 움직임에 대한 정보를 묘사할 수 있게 하였다. 매개변수화 된 움직임 벡터는 눈 영역, 입술과 눈썹 영역, 그리고 얼굴영역을 위한 서로 다른 세 종류의 움직임을 위하여 사용하였다. 이를 얼굴 모델의 움직임을 합성할 수 있는 단위행위(Action Unit)와 결합하여 2차원 동영상에서의 얼굴 움직임을 3 차원으로 합성한 결과를 얻을 수 있다.

• 한국의류산업학회지
• /
• 제7권5호
• /
• pp.527-534
• /
• 2005

The purpose of this research is to study face images according to color of make-up was made by computer graphic simulation. The various facial images can be helpful for choosing suitable make-up color planning. In order to find out the differences of face images by make-up color, three different foundations and seven eye-shadows, six lips were applied on the round face model. Make-up Image Scale was used the scale of seven point modified the S-D method. Data were analyzed by Varimax perpendicular rotation method, Duncan's Multiple Range Test, Three-way ANOVA. As the result of make-up image perception analysis, a factor structure was divided into mildness, modernness, elegance, unique. The factor of mildness, modernness, unique affected on the foundation color. Foundation color was found out to be influential variable to distinguish color perception abilities. Also, the foundation, eye-shadow, lip color were influenced interactively on the perception of elegance factor. Pink color was important color, influenced on the mildness factor. Gray and purple color were influenced on the modernness factor. Mildness factor was perceived as the most bright foundation but unique factor was perceived as the most dark foundation. Then, the foundation, eye-shadow, lip color were influenced interactively on the perception of facial images. The results can be effectively applied to today's marketing and color design management which is focused on the product's emotional image in customer's mind.

• 대한구순구개열학회지
• /
• 제4권1호
• /
• pp.1-11
• /
• 2001

Disposable blade is widely used for palatal and oral mucosal incision in oral and maxillofadal surgery nowadays, But its design and durability need for improvement, Especially, there are so many hard tissues intraoral area, such as bone and tooth, therefor the sharpness of the surgical blade was easily destroyed, The purpose of this study was to make basic data for developing new design of surgical blade using in oral and maxillofacial area including for the patients who have cleft lip and palate deformities, Some questionnaires about the usefulness of currently used surgical blades were sent to 150 dentists, the 54 of them made a reply, Secondly, The used-once blade and fresh new blade were examined under the scanning electron microscope with the 4000-times magnification, Lastly, the tissue reaction following the surgical incision with a fresh-new and a used blade on rat buccal cheek mucosa and hard palate was evaluated with light microscope with hematoxilin-eosin staining, The time interval from the surgical trauma to taking a sample were 1 day, 3 days, 7 days, and 14 days, At each time schedule, 2 Sprague-Dawley rats were sacrificed, Many dentists were agreed to need for changing the design of the surgical blades and also demand to improve the durability of the blades, They were also eager to adopt the new design of blade if it was available, The blade used in surgical extraction procedure was heavily damaged in its sharpe edge of number 15 blade, The histological differences were not prominent, but the delayed healing was detected in buccal mucosal defects especially in the surgical group with used blade, There are slight different changes in hard palatal defects between a used and a new blade group, In this study, we could find that there are imperative demanding on improvement of surgical blade design and durability for oral and maxillofadal area, The blade currently using in surgical extraction was easily damaged, The animal model of this study was not perfect for the purpose of this study.

• 대한구순구개열학회지
• /
• 제4권1호
• /
• pp.13-26
• /
• 2001

Cleft lip and/or palate is the congenital orofacial malformation most commonly occurred in humans, The disease is multifactorial and is probably caused by genetic and/or environmental factors, So, there are many problems in research concerning etiology and in treatment of the disease, Even the most practiced and sophisticated methods of surgical repair are necessarily followed by scar contraction and fibrosis, which result in skeletal defects, dental abnormalities, cosmetic disfigurement, and speech impairment, As a result, Fetal surgery can be considered but practiced rarely when the deformity is not fatal to life, And treatment of cleft palate is performed in the form of medicine projection into uterus in animal experiments, Many studies show that growth factor and its receptor emerge from the developing palate; and the epidermal growth factor receptors have a important role in craniofacial development and in palatal fusion, The palatal morphogenesis of the avine is different from the mammal's; it takes the form of physiologic cleft palate, Recently, cleft palate fusion experiment was performed when the avine were in the period of palate formation through the exogenous TGF-β3 addition, and it showed that the exogenous TGF-β3 makes fusion of divided palate through certain process when cleft palate is occurred in palatal formation, In this study, I had the conformation of the fusion of cleft palate through the addition of TGF-β in case of chicken embryo, and observed the effect of TGF-β in EGF receptor distribution, And the following is the results of this study, 1. In case of the TGF-βl and β3 addition group, there was the decrease of EGFR(Epidermal Growth Factor Receptor) immunoreactivity in mesenchymal cells beneath the medial edge epithelium and also in epithelial mesenchymal interface which is between medial edge epithelium and nasal septum in 72 hours, 2, The immunoreactivity of the control group resembles that of normal chicken embryo palate in development, 3. In the view through fluorescence confocal microscopy, there was confluence in TGF-β3 addition group, This shows that the confluence induced by exogenous TGF-β3 is related to EGFR expression in palate of chicken embryo, which is a physiologic cleft palate model.

• 한국윤활학회:학술대회논문집
• /
• 한국윤활학회 2002년도 proceedings of the second asia international conference on tribology
• /
• pp.455-456
• /
• 2002

Wear is one of the main factors causing breakdown and fault of machine, so ferrography technique analyzing wear particles can be an effective way for condition monitoring and fault diagnosis. On the base of the forward multilayer neural network, a nodes self-deleting neural network model is provided in this paper. This network can itself deletes the nodes to optimize its construction. On the basis of the nodes self-deleting neural network, an intelligent ferrography diagnosis expert system (IFDES) for wear particles recognition and wear diagnosis is described. This intelligent expert system can automatically slim lip knowledge by learning from samples and realize basically the entirely automatic processing from wear particles recognition to wear diagnosis.

• Advances in robotics research
• /
• 제2권2호
• /
• pp.151-159
• /
• 2018

The evolution of bipedal robots was the foundation stone for development of Humanoid robots. The highly complex and non-linear dynamic of human walking made it very difficult for researchers to simulate the gait patterns under different conditions. Simple controllers were developed initially using basic mechanics like Linear Inverted Pendulum (LIP) model and later on advanced into complex control systems with dynamic stability with the help of high accuracy feedback systems and efficient real-time optimization algorithms. This paper illustrates a number of significant mathematical models and controllers developed so far in the field of bipeds and humanoids. The key facts and ideas are extracted and categorized in order to describe it in a comprehensible structure.

• 대한음성학회지:말소리
• /
• 제56호
• /
• pp.159-172
• /
• 2005

This paper analyzes the vocal tract area estimation algorithm used as a part of a speech analysis program to help deaf children correct their pronunciations by comparing their vocal tract shape with normal children's. Assuming that a vocal tract is a concatenation of cylinder tubes with a different cross section, we compute the relative vocal tract area of each tube using the reflection coefficients obtained from linear predictive coding. Then, we obtain the absolute vocal tract area by computing the height of lip opening with a formula modified for children's speech. Using the speech data for five Korean vowels (/a/, /e/, /i/, /o/, and /u/), we investigate the effects of the sampling frequency, frame size, and model order on the estimated vocal tract shape. We compare the vocal tract shapes obtained from deaf and normal children's speech.

• 한국정보과학회:학술대회논문집
• /
• 한국정보과학회 2001년도 가을 학술발표논문집 Vol.28 No.2 (2)
• /
• pp.412-414
• /
• 2001

In this paper, we describe an real-time facial feature tracker. We only used a general USB PC Camera without a frame grabber. The system has achieved a rate of 8+ frames/second without any low-level library support. It tracks pupils, nostrils and corners of the lip. The signal from USB Camera is YUV 4:2:0 vertical Format. we converted the signal into RGB color model to display the image and We interpolated V channel of the signal to be used for extracting a facial region. and we analysis 2D blob features in the Y channel, the luminance of the image with geometric restriction to locate each facial feature within the detected facial region. Our method is so simple and intuitive that we can make the system work in real-time.