• 재활복지공학회논문지
• /
• 제8권3호
• /
• pp.145-150
• /
• 2014

본 논문에서는 입력된 신호에 의해 결정되는 각 주파수 밴드별 우도비(likelihood ratio, LR)를 deep belief networks(DBN)의 입력층으로 이용하는 새로운 음성 검출기(voice activity detection, VAD) 알고리즘을 제안한다. 기존의 통계적 모델 기반의 음성 검출기는 음성 구간을 판단하기 위해 우도비를 기하 평균을 이용한 결정식을 사용한다. 제안된 음성 검출기는 이 결정식을 대신해 DBN을 이용하여, 오검출 확률을 최소화 하도록 학습을 한다. 제안된 DBN 기반의 음성 검출 알고리즘은 통계적 모델 기반의 음성 검출기의 성능을 개선한 support vector machine(SVM) 기반의 음성 검출기와 정상 및 비정상 잡음 환경에서 다양한 조건을 부과하여 비교하였다. 제안된 알고리즘이 기존의 SVM 기반의 알고리즘보다 전체 오분류 확률 [0.7, 2.7]의 향상 폭을 보였다.

• 인터넷정보학회논문지
• /
• 제7권3호
• /
• pp.143-153
• /
• 2006

얼굴 특징 검출은 HCI, 얼굴 인식, 얼굴 추적, 표정 인식 및 이미지 데이터 검색등과 같은 응용분야의 근간 기술이다. 실시간 환경에서 얼굴 특징 검출을 처리하기 위해서는 검출 알고리즘의 속도가 중요한 관건으로 작용하고 있다. 또한 빛의 변화, 대상의 위치, 각도, 복잡한 배경등과 같은 요인들은 얼굴 특징 검출 알고리즘의 검출율을 낮추는데 영향을 미치므로 이를 개선한 방법이 필요하다. 본 연구에서는 검출율과 검출 속도를 동시에 개선한 알고리즘을 제안한다. 제안 알고리즘은 얼굴 이미지에 빛 보상 알고리즘인 CLAHE를 이용하여 빛의 변화에 강건하도록 이미지를 개선한 다음 얼굴 피부 영역을 검출한다. 검출한 피부 영역에서 얼굴 특징 포인트를 추출하기 위해 얼굴 특징의 외형기반 기하학적 성질을 이용한다. 제안 알고리즘은 얼굴 특징 검출의 정확도를 높일 뿐 아니라 빠른 검출 속도를 보임으로써 얼굴 추적, 인식 등과 같은 실시간 응용분야에 적용할 수 있다.

• 한국동물위생학회지
• /
• 제39권1호
• /
• pp.35-39
• /
• 2016

The purpose of this study was to survey on infection status of pathogens of diarrhea from Korean indigenous goat. A total of 800 fecal samples was collected from 50 farms from January to October 2015 and was tested by automatic biochemical machine and polymerase chain reaction (PCR). The overall detection ratio of bacterial pathogens was 22.4% and viral pathogens was 16.3%, respectively. The detection ratio of Escherichia coli (E. coli), Salmonella spp., bovine viral diarrhea virus (BVDV), rotavirus and coronavirus were 21.5%, 0.9%, 7.6%, 5.6% and 3.0%, respectively. In the rates of mixed detection, single was 78.2%, double 8.4%, triple 11.6% and quadruple 1.8% in each sample and 38%, 12%, 16%, 20% in each farm, respectively.

• 산업기술연구
• /
• 제22권A호
• /
• pp.29-36
• /
• 2002

A damage detection algorithm using vibration modes is applied to the ring structures and the modal behaviors of the slightly asymmetric rings are examined. Mode shape change, MSER(modal strain energy ratio) and MCR(modal curvature ratio) are investigated to identify the locations of faults or damages The above fault detection parameters are calculated and compared by the finite element analysis on rings with designed local damages. Damages are modeled as a reduced stiffness in the analysis The results show that MSER and MCR can be proper parameters to detect local damages in the ring structures.

• International journal of advanced smart convergence
• /
• 제12권1호
• /
• pp.64-69
• /
• 2023

We devise a collaboration construction method based on the SPRT (Sequential Probability Ratio Test) for malware detection in IoT. In our method, high-end IoT nodes having capable of detecting malware and generating malware signatures harness the SPRT to give a reward of malware signatures to low-end IoT nodes providing useful data for malware detection in IoT. We evaluate our proposed method through simulation. Our simulation results indicate that the number of malware signatures provided for collaboration is varied in accordance with the threshold for fraction of useful data.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• 제15권4호
• /
• pp.251-259
• /
• 2015

In this paper, a ship detection method is proposed; this method uses edge-based segmentation and histogram of oriented gradient (HOG) with the ship size ratio. The proposed method can prevent a marine collision accident by detecting ships at close range. Furthermore, unlike radar, the method can detect ships that have small size and absorb radio waves because it involves the use of a vision-based system. This system performs three operations. First, the foreground is separated from the background and candidates are detected using Sobel edge detection and morphological operations in the edge-based segmentation part. Second, features are extracted by employing HOG descriptors with the ship size ratio from the detected candidate. Finally, a support vector machine (SVM) verifies whether the candidates are ships. The performance of these methods is demonstrated by comparing their results with the results of other segmentation methods using eight-fold cross validation for the experimental results.

• 대한음성학회지:말소리
• /
• 제59호
• /
• pp.27-51
• /
• 2006

For speech recognition systems, rejection function as well as decoding function is necessary to improve the reliability. There have been many research efforts on out-of-vocabulary word rejection, however, little attention has been paid on non-target sentence rejection. Recently pronunciation approaches using speech recognition increase the need for non-target sentence rejection to provide more accurate and robust results. In this paper, we proposed filler model method and word/phoneme detection ratio method to implement non-target sentence rejection system. We made performance evaluation of filler model along to word-level, phoneme-level, and sentence-level filler models respectively. We also perform the similar experiment using word-level and phoneme-level word/phoneme detection ratio method. For the performance evaluation, the minimized average of FAR and FRR is used for comparing the effectiveness of each method along with the number of words of given sentences. From the experimental results, we got to know that word-level method outperforms the other methods, and word-level filler mode shows slightly better results than that of word detection ratio method.

• 제어로봇시스템학회논문지
• /
• 제22권8호
• /
• pp.665-670
• /
• 2016

This paper proposes a repeat-back jamming signal detection scheme using a correlation ratio test metric of a combined pseudo-random noise signal for global navigation satellite systems. The correlation ratio test metric allows for the monitoring of possible distortions in the signal correlation. The proposed scheme is a modified version of the correlation ratio test metric to detect a repeat-back jamming signal in a multipath environment. Through a Monte-Carlo simulation, it is confirmed that the proposed scheme detects almost the whole case, which is received a repeat-back jamming signal under the 6 dB jamming to signal power ratio.

• 대한원격탐사학회지
• /
• 제26권3호
• /
• pp.287-295
• /
• 2010

Water level of the Tonle Sap is largely influenced by the Mekong River. During the wet season, the lacustrine landform and vegetated areas are covered with water. Change detection in this area provides information required for human activities and sustainable development around the Tonle Sap. In order to detect the changes in the Tonle Sap floodplain, fifteen ALOS-PALSAR L-band data acquired from January 2007 to January 2009 and examined in this study. Since L-band is able to penetrate into vegetation cover, it enables us to study the changes according to water level of floodplain developed in the rainforest. Four types of images were constructed and studied include 1) ratio images, 2) correlation coefficient images, 3) texture feature ratio images and 4) multi-color composite images. Change images (in each 46 day interval) extracted from the ratio images, coherence images and texture feature ratio images were formed for detecting land cover change. Two RGB images are also obtained by compositing three images acquired in the early, in the middle and at the end of the rainy season in 2007 and 2008. Combination of the methods results that the change images present the relationship between vegetation and water level, leaf fall forest as well as cultivation and harvest crop.

• Nuclear Engineering and Technology
• /
• 제53권9호
• /
• pp.3122-3125
• /
• 2021

Electromagnetic vortex flowmeter is a new type of instrument for detecting leakage of steam generator, and the signal processing method based on the envelope to calculate energy ratio can effectively detect bubbles in sodium flow. The signal processing method is not affected by changes in the amplitude of the sensor output signal, which is caused by changes in magnetic field strength and other factors. However, the detection sensitivity of the electromagnetic vortex flowmeter is reduced. To this end, a signal processing method based on inverse Fourier transform to calculate energy ratio is proposed. According to the difference between the frequency band of the bubble noise signal and the flow signal, only the amplitude in the frequency band of the flow signal is retained in the frequency domain, and then the flow signal is obtained by the inverse Fourier transform method, thereby calculating the energy ratio. Using this method to process the experimental data, the results show that it can detect 0.1 g/s leak rate of water in the steam generator, and its performance is significantly better than that of the signal processing method based on the envelope to calculate energy ratio.