• The Journal of the Acoustical Society of Korea
• /
• v.15 no.6
• /
• pp.73-80
• /
• 1996

Success of a face recognition system depends on which parameters are used. Generally the parameters are affected by environment of facial images such as illumination. To reduce the influence of the evcironment, since side images are insensitive to variance of brightness, it might be an appropriate approach to make the defect of front face images complete with the features extracted from side images. This paper proposes a method which collects and completes the information of front and side images. It is intended to prove the usefulness of the method that it is compared with other methods.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.45 no.4
• /
• pp.92-99
• /
• 2008

We propose an effective voice-pishing detection algorithm based on the 3GPP2 selectable mode vocoder (SMV). The detection of voice pishing is performed based on a Gaussian mixture model (GMM) using decoding parameters of the SMV directly extracted from the decoding process of the transmitted speech information in the mobile phone. The experimental results indicate that SMV decoding parameters are effective in discriminating between general voice and phisher's voice and the performance is significantly acceptable when the proposed technique is applied.

• The Journal of the Acoustical Society of Korea
• /
• v.26 no.5
• /
• pp.207-213
• /
• 2007

In this research, we propose an efficient Speech/Music discrimination method that uses spectrum analysis and neural network. The proposed method extracts the duration feature parameter(MSDF) from a spectral peak track by analyzing the spectrum, and it was used as a feature for Speech/Music discriminator combined with the MFSC. The neural network was used as a Speech/Music discriminator, and we have reformed various experiments to evaluate the proposed method according to the training pattern selection, size and neural network architecture. From the results of Speech/Music discrimination, we found performance improvement and stability according to the training pattern selection and model composition in comparison to previous method. The MSDF and MFSC are used as a feature parameter which is over 50 seconds of training pattern, a discrimination rate of 94.97% for speech and 92.38% for music. Finally, we have achieved performance improvement 1.25% for speech and 1.69% for music compares to the use of MFSC.

• Journal of the Institute of Convergence Signal Processing
• /
• v.20 no.2
• /
• pp.70-77
• /
• 2019

Breast ultrasound readings are very important to diagnose early breast cancer. In Ultrasonic inspection, it shows a significant difference in image quality depending on the ultrasonic equipment, and there is a large difference in diagnosis depending on the experience and skill of the inspector. Therefore, objective criteria are needed for accurate diagnosis and treatment. In this study, we analyzed texture characteristics by applying GLCM (Gray Level Co-occurrence Matrix) algorithm and extracted characteristic parameters and diagnosed breast cancer using neural network classifier. Breast ultrasound images were classified into normal, benign and malignant tumors and six texture parameters were extracted. Fourteen cases of normal, malignant and benign tumor diagnosed by mammography were studied by using the extracted six parameters and learning by multi - layer perceptron neural network back propagation learning method. As a result of classification using 51 normal images, 62 benign tumor images, and 74 malignant tumor images of the learned model, the classification rate was 95.2%.

• Proceedings of the Korean Information Science Society Conference
• /
• 1998.10c
• /
• pp.429-431
• /
• 1998

본 논문에서는 의료 영상 중에서 성대 운동의 불규칙적인 움직임을 판단하여 자동으로 진단 파라미터를 구하는 비디오스트로보키모그래피(Videostrobokymography) 시스템에서 관심 영역을 추출하는 방법을 소개하고자 한다. CCD카메라에 의해 촬영된 영상은 비디오 테이프에 저장된 후 이미지 캡쳐 보드에서 그레이 이미지(gray-level)로 변환되어 저장된다. 입력된 영상은 움직이는 영상을 촬영한 것이므로 관심 영역의 위치가 각 프레임마다 다르다. 또한 실제로 입력된 성대영상들이 점진적인 농도 변화를 보이기 때문에 에지에 의해 영역을 추출하는 일반적인 영역 추출방법은 사용하기 어렵다. 본 논문에서는 두 번의 단계를 통하여 관심 영역을 추출하고 있다. 첫 번째는 입력된 영상에서 노이즈를 제거한 후 각 프레임에서 영상의 최소 에너지를 구한다. 두 번째로 농도 변화 값을 특징 값으로 이용하는 분할-합병 알고리즘(Split-merge Algorithm)을 적용하여 관심 영역을 추출하였다. 제안한 알고리즘을 19명의 성대 영상에 적용하여 분석한 결과 성대의 관심 영역을 추출할 수 있었다. 그리고, 영상의 에너지 값을 이용하는 스네이크 알고리즘(Snake Algorithm)에 적용하여 비교해본 결과 본 연구에서 제안하는 스네이크 알고리즘보다 좋은 성능을 보임을 확인할 수 있었다. 본 연구에서 제안하는 관심 영역 추출 방법은 동적인 변화를 보이는 영상에서 관심 영역을 추출할 수 있을 뿐 아니라 계산 량이 적어 200x280크기의 이미지를 초당 약 40프레임에 대한 관심 영역을 추출할 수 있는 장점이 있다.

• Journal of the Korea Society of Computer and Information
• /
• v.10 no.6 s.38
• /
• pp.47-54
• /
• 2005

In the most of speaker recognition system, speaker's characteristics is extracted from acoustic parameter by speech analysis and we make speaker's reference pattern. Parameters used in speaker recognition system are desirable expressing speaker's characteristics fully and being a few difference whenever it is spoken. Therefore we su99est following to solve this problem. This paper is proposed to use strong spectrum characteristic in non-noise circumstance and prosodic information in noise circumstance. In a stage of making code book, we make the number of data we need to combine spectrum characteristic and Prosodic information. We decide acceptance or rejection comparing test pattern and each model distance. As a result, we obtained more improved recognition rate than we use spectrum and prosodic information especially we obtained stational recognition rate in noise circumstance.

• Journal of Internet Computing and Services
• /
• v.22 no.2
• /
• pp.1-9
• /
• 2021

Traditionally, most malicious codes have been analyzed using feature information extracted by domain experts. However, this feature-based analysis method depends on the analyst's capabilities and has limitations in detecting variant malicious codes that have modified existing malicious codes. In this study, we propose a ResNet-Variational AutoEncder-based variant malware classification method that can classify a family of variant malware without domain expert intervention. The Variational AutoEncoder network has the characteristics of creating new data within a normal distribution and understanding the characteristics of the data well in the learning process of training data provided as input values. In this study, important features of malicious code could be extracted by extracting latent variables in the learning process of Variational AutoEncoder. In addition, transfer learning was performed to better learn the characteristics of the training data and increase the efficiency of learning. The learning parameters of the ResNet-152 model pre-trained with the ImageNet Dataset were transferred to the learning parameters of the Encoder Network. The ResNet-Variational AutoEncoder that performed transfer learning showed higher performance than the existing Variational AutoEncoder and provided learning efficiency. Meanwhile, an ensemble model, Stacking Classifier, was used as a method for classifying variant malicious codes. As a result of learning the Stacking Classifier based on the characteristic data of the variant malware extracted by the Encoder Network of the ResNet-VAE model, an accuracy of 98.66% and an F1-Score of 98.68 were obtained.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.12 no.4
• /
• pp.349-358
• /
• 2019

This paper aimed at collecting sensor signals to extract characteristic parameter of the rotor. A vibration test rig has been developed to perform model tests. Signal characteristics were analyzed when driving normally. Envelope FFT Analysis is used to extract vibration components caused by periodic impacts from other vibration factors. Signal analysis was performed when load changes were given to speed sensors and vibration test rigs that show low frequency characteristics of the rotor and signal analysis according to rotational speed. The acceleration signal measured in the bearing housing has a small amplitude and produces only the rotational frequency component and harmonic component of the motor. As the number of rotations increases, the amplitude of acceleration can be seen. As the rotational speed increases, it can be seen that there is a difference in the shape of the original data and compared with the acceleration FFT graph, it can be seen that the noise is strong at low frequencies and the corresponding rotational frequency components are clearly represented. It can be seen that changing the load does not increase the main rotational frequency component.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.21 no.6
• /
• pp.679-686
• /
• 2020

In this thesis, an enhanced method for the feature extraction of vocal source signals and score combination using an MCE-Based weight estimation of the score of multiple feature vectors are proposed for the performance improvement of speaker recognition systems. The proposed feature vector is composed of perceptual linear predictive cepstral coefficients, skewness, and kurtosis extracted with lowpass filtered glottal flow signals to eliminate the flat spectrum region, which is a meaningless information section. The proposed feature was used to improve the conventional speaker recognition system utilizing the mel-frequency cepstral coefficients and the perceptual linear predictive cepstral coefficients extracted with the speech signals and Gaussian mixture models. In addition, to increase the reliability of the estimated scores, instead of estimating the weight using the probability distribution of the convectional score, the scores evaluated by the conventional vocal tract, and the proposed feature are fused by the MCE-Based score combination method to find the optimal speaker. The experimental results showed that the proposed feature vectors contained valid information to recognize the speaker. In addition, when speaker recognition is performed by combining the MCE-based multiple feature parameter scores, the recognition system outperformed the conventional one, particularly in low Gaussian mixture cases.

• Journal of the Korean Society of Radiology
• /
• v.15 no.4
• /
• pp.473-479
• /
• 2021

The diagnostic criteria for diffuse thyroid disease are ambiguous and there are many errors due to the subjective diagnosis of experts. Also, studies on ultrasound imaging of thyroid nodules have been actively conducted, but studies on diffuse thyroid disease are insufficient. In this study, features were extracted by applying the GLCM algorithm to ultrasound images of normal and diffuse thyroid disease, and quantitative analysis was performed using the extracted feature values. Using the GLCM algorithm for thyroid ultrasound images of patients diagnosed at W hospital, 199 normal cases, 132 mild cases, and 99 moderate cases, a region of interest (50×50 pixel) was set for a total of 430 images, and Autocorrelation, Sum of squares, sum average, sum variance, cluster prominence, and energy were analyzed using six parameters. As a result, in autocorrelation, sum of squares, sum average, and sum variance four parameters, Normal, Mild, and Moderate were distinguished with a high recognition rate of over 90%. This study is valuable as a criterion for classifying the severity of diffuse thyroid disease in ultrasound images using the GLCM algorithm. By applying these parameters, it is expected that errors due to visual reading can be reduced in the diagnosis of thyroid disease and can be utilized as a secondary means of diagnosing diffuse thyroid disease.