• The Journal of the Acoustical Society of Korea
• /
• v.39 no.5
• /
• pp.441-446
• /
• 2020

In this paper, we propose the α-Feature Map Scaling (α-FMS) method which extends the FMS method that was designed to enhance the discriminative power of feature maps of deep neural networks in Speaker Verification (SV) systems. The FMS derives a scale vector from a feature map and then adds or multiplies them to the features, or sequentially apply both operations. However, the FMS method not only uses an identical scale vector for both addition and multiplication, but also has a limitation that it can only add a value between zero and one in case of addition. In this study, to overcome these limitations, we propose α-FMS to add a trainable parameter α to the feature map element-wise, and then multiply a scale vector. We compare the performance of the two methods: the one where α is a scalar, and the other where it is a vector. Both α-FMS methods are applied after each residual block of the deep neural network. The proposed system using the α-FMS methods are trained using the RawNet2 and tested using the VoxCeleb1 evaluation set. The result demonstrates an equal error rate of 2.47 % and 2.31 % for the two α-FMS methods respectively.

• Journal of the Institute of Electronics Engineers of Korea TC
• /
• v.43 no.7 s.349
• /
• pp.32-40
• /
• 2006

It is very important to establish a pairwise key securely in wireless sensor networks. Because sensor networks consist of devices with weak physical security, they are likely to be compromised by an attacker. However, some approaches using key pre-distribution and other approaches using one hop local keys are known to be very vulnerable to threats caused by compromised nodes, even a small number. This paper proposes a scheme where each node establishes three hop local keys and employs them for a later pairwise key establishment. When any two nodes agree a pairwise key, all nodes on the route between two nodes contribute to the agreement of the pairwise key. Here, the initial three hop local keys are employed for encrypting a secret key delivered from a node to other nodes. Therefore, the proposed scheme bothers attackers to compromise much more nodes than the scheme using one hop local keys only. The simulation results have proven that the proposed scheme provides better performance and higher security than the scheme using one hop local keys in terms of message exchange, the number of encryption and decryption, and pairwise key exposure rate.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.45 no.1
• /
• pp.11-21
• /
• 2008

In conventional video coding, the complexity of encoder is much higher than that of decoder. However, as more needs arises for extremely simple encoder in environments having constrained energy such as sensor network, much investigation has been carried out for eliminating motion prediction/compensation claiming most complexity and energy in encoder. The Wyner-Ziv coding, one of the representative schemes for the problem, reconstructs video at decoder by correcting noise on side information using channel coding technique such as turbo code. Since the encoder generates only parity bits without performing any type of processes extracting correlation information between frames, it has an extremely simple structure. However, turbo decoding errors occur in noisy side information. When there are high-motion or occlusion between frames, more turbo decoding errors appear in reconstructed frame and look like Salt & Pepper noise. This severely deteriorates subjective video quality even though such noise rarely occurs. In this paper, we propose a computationally extremely light encoder based on symbol-level Wyner-Ziv coding technique and a new corresponding decoder which, based on a decision whether a pixel has error or not, applies median filter selectively in order to minimize loss of texture detail from filtering. The proposed method claims extremely low encoder complexity and shows improvements both in subjective quality and PSNR. Our experiments have verified average PSNR gain of up to 0.8dB.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.31 no.3A
• /
• pp.368-376
• /
• 2006

This paper describes a Fcore_GenSim(Parameterized FFT Core Generation & Simulation Program), which can be used as an essential If(Intellectual Property) in various OFDM modem designs. The Fcore_Gensim is composed of two parts, a parameterized core generator(PFFT_CoreGen) that generates Verilog-HDL models of FFT cores, and a fixed-point FFT simulator(FXP_FFTSim) which can be used to estimate the SQNR performance of the generated cores. The parameters that can be specified for core generation are FFT length in the range of 64 ~2048-point and word-lengths of input/output/internal/twiddle data in the range of 8-b "24-b with 2-b step. Total 43,659 FFT cores can be generated by Fcore_Gensim. In addition, CBFP(Convergent Block Floating Point) scaling can be optionally specified. To achieve an optimized hardware and SQNR performance of the generated core, a hybrid structure of R2SDF and R2SDC stages and a hybrid algorithm of radix-2, radix-2/4, radix-2/4/8 are adopted according to FFT length and CBFP scaling.

• Journal of Intelligence and Information Systems
• /
• v.15 no.2
• /
• pp.33-48
• /
• 2009

Capsule endoscopy is one of the most remarkable inventions in last ten years. Causing less pain for patients, diagnosis for entire digestive system has been considered as a most convenience method over a normal endoscope. However, it is known that the diagnosis process typically requires very long inspection time for clinical experts because of considerably many duplicate images of same areas in human digestive system due to uncontrollable movement of a capsule endoscope. In this paper, we propose a method for clinical diagnosticians to get highly valuable information from capsule-endoscopy video. Our software system consists of three global maps, such as movement map, characteristic map, and brightness map, in temporal domain for entire sequence of the input video. The movement map can be used for effectively removing duplicated adjacent images. The characteristic and brightness maps provide frame content analyses that can be quickly used for segmenting regions or locating some features(such as blood) in the stream. Our experiments show the results of four patients having different health conditions. The result maps clearly capture the movements and characteristics from the image frames. Our method may help the diagnosticians quickly search the locations of lesion, bleeding, or some other interesting areas.

• Journal of Broadcast Engineering
• /
• v.10 no.3
• /
• pp.270-285
• /
• 2005

A joint design scheme is proposed to optimize the up/down scaler and the motion vector estimation module in the transcoder system. The proposed scheme first optimizes the resolution scaler for a fixed motion vector, and then a new motion vector is estimated for the fixed scaler. These two steps are iteratively repeated until they reach a local optimum solution. In the optimization of the scaler, we derive an adaptive version of a cubic convolution interpolator to enlarge or reduce digital images by arbitrary scaling factors. The adaptation is performed at each macroblock of an image. In order to estimate the optimal motion vector, a temporary motion vector is composed from the given motion vectors. Then the motion vector is refined over a narrow search range. It is well-known that this refinement scheme provides the comparable performance compared to the full search method. Simulation results show that a jointly optimized system based on the proposed algorithms outperforms the conventional systems. We can also see that the algorithms exhibit significant improvement in the minimization of information loss compared with other techniques.

• Journal of KIISE:Computing Practices and Letters
• /
• v.10 no.6
• /
• pp.514-528
• /
• 2004

An emotion-based video scene retrieval algorithm is proposed in this paper. First, abrupt/gradual shot boundaries are detected in the video clip representing a specific story Then, five video features such as 'average color histogram' 'average brightness', 'average edge histogram', 'average shot duration', and 'gradual change rate' are extracted from each of the videos and mapping between these features and the emotional space that user has in mind is achieved by an interactive genetic algorithm. Once the proposed algorithm has selected videos that contain the corresponding emotion from initial population of videos, feature vectors from the selected videos are regarded as chromosomes and a genetic crossover is applied over them. Next, new chromosomes after crossover and feature vectors in the database videos are compared based on the similarity function to obtain the most similar videos as solutions of the next generation. By iterating above procedures, new population of videos that user has in mind are retrieved. In order to show the validity of the proposed method, six example categories such as 'action', 'excitement', 'suspense', 'quietness', 'relaxation', 'happiness' are used as emotions for experiments. Over 300 commercial videos, retrieval results show 70% effectiveness in average.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2015.05a
• /
• pp.590-590
• /
• 2015

본 연구는 집중호우에 의한 홍수예측 및 소유역의 유출거동에 대한 수문학적 민감성(susceptibility) 규명을 목적으로 강우강도, 지속기간 및 토양포화도 변화에 따른 홍수유출특성을 분석하여 유역의 유출거동 민감성을 표출할 수 있는 노모그래프를 개발하였다. 개별 홍수사상에 대한 유출거동 특성 분석을 위하여 한국건설기술연구원의 대표 시험유역인 설마천 유역의 과거 17년간(1996 ~ 2012)의 10분 간격의 강우량 및 유출량 자료를 수집하여 홍수유출해석을 수행하였다. 설마천 시험유역의 일누가강우량 100mm 이상, 50개 홍수사상에 대한 홍수유출해석은 유역 물순환 해석모형인 CAT(Catchment hydrological cycle Assessment Tool)을 이용하였으며 모의결과를 바탕으로 홍수사상별 지체시간, 강우강도, 지속기간 및 토양포화도 변화에 따른 홍수유출특성을 상세히 분석하였다. 이 중에서도 지체시간은 유역반응을 나타내는 시간변수로서 수문모델링 및 홍수량예측에 매우 중요한 요소이다. 특히, 강우량에 대한 홍수량의 반응이 빠른 소유역의 경우에 홍수량예측에 큰 영향을 미친다. 따라서 강우강도, 지속기간, 토양포화도의 변화량에 대한 지체시간의 거동을 R 프로그래밍 언어 및 3D Surfer를 이용하여 분석한 후 최종적으로 소유역의 홍수유출 특성을 나타내는 3차원 홍수 유출특성 노모그래프를 개발하였다. 분석에 사용된 R 프로그래밍 언어는 통계 계산과 그래픽을 위한 프로그래밍 언어이자 소프트웨어 환경으로 데이터의 조작 및 수치연산, 시각화를 수행할 수 있는 기능을 여러 패키지를 통해 구현할 수 있다. 따라서 본 연구에서는 R을 이용하여 10분 단위의 강우 및 유출량 자료를 1시간 및 1일 자료로 구축하고 17년간의 과거 홍수사상을 분리하여 추출하는 R 홍수유출해석 시스템을 개발하였으며 추출된 홍수사상을 관측 유출량 및 관측 토양수분을 포함하여 시각화함으로써 강우 및 토양수분 변화에 따른 소유역의 유출거동 민감성을 확인할 수 있었다. 분석 결과, 지체시간은 강우지속기간 및 토양포화도에 민감한 거동특성을 나타냈으며 토양포화도는 첨두홍수량의 변화에 민감한 영향을 주는 것으로 확인되었다. 개발된 3차원 홍수유출특성 노모그래프는 유역의 규모 및 지형물리학적 특성에 따라 다양하게 나타날 것으로 판단되며 여러 계측유역에 적용함으로써 유역별 홍수유출 반응특성을 정량화할 필요가 있다. 즉, 강우강도, 지속기간, 지체시간, 포화도 등의 변화에 따른 유역의 홍수유출 반응특성을 규명함으로써 미계측 유역의 홍수량예측 실무에 활용할 수 있을 것으로 판단된다.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.30 no.2B
• /
• pp.219-231
• /
• 2010

The Reference Evapotranspiration Calculator Software (REF-ET) supports computational guidelines for the reference evapotranspiration using seventeen FAO Penman-Monteith (PM) equations simultaneously such as the ASCE and FAO standardized forms. The REF-ET can conveniently consider missing data predictions and regional site characterizations, when reference ET is computed on monthly, daily, and hourly time steps. The applicability of the REF-ET was estimated to simulate the reference ET using hourly weather data from Seoul weather station for 29 years. The result found that the FAO24-Rd and 1957-Makk equations closely concerned with solar radiation parameter which were the most highly correlated to reference ET computed by pan coefficient. In addition, the 1957-Makk equation was identified as the most correct computational method for reference ET by analysis of bias and root mean square error. The 1957-Makk equation could predict the reference ET within the error of less than 1.06 mm/day, though all the other equations tended toward overestimation of predicting the reference ET in comparison with refecence ET of pan. The results of this study suggest that the REF-ET will be applicable to support reference ET estimation for a variety of field condition and time-scale.

• Journal of the Korean Recycled Construction Resources Institute
• /
• v.11 no.4
• /
• pp.355-363
• /
• 2023

In safety assessment of a aged bridge, dynamic characteristics and displacement are directly related to the rigidity of the structural system, especially displacement is the most important factor as the physical quantity that the bridge user can directly detect. However, in order to measure the displacement of the bridge, it is difficult to install displacement sensors at the bottom of the bridge and conduct traffic blocking and loading tests, resulting in increased costs or impossible measurements depending on the bridge's environment. In this study, a method of measuring the displacement of a bridge using only accelerometers without installing displacement sensors and ambient vibration without a loading test was proposed. For the analysis of bridge dynamic characteristics and displacement using ambient vibration, the mode shape and natural frequency of the bridge were extracted using a TDD technique known to enable quick analysis with simple calculations, and the unit load displacement of the bridge was analyzed through flexibility analysis to calculate static displacement. To verify this proposed technology, an on-site test was conducted on C Bridge, and the results were compared with the measured values of the loading test and the structural analysis data. As a result, it was confirmed that the mode shape and natural frequency were 0.42 to 1.13 % error ratio, and the maximum displacement at the main span was 3.58 % error ratio. Therefore, the proposed technology can be used as a basis data for indirectly determine the safety of the bridge by comparing the amount of displacement compared to the design and analysis values by estimating the displacement of the bridge that could not be measured due to the difficulty of installing displacement sensors.