• Journal of the Institute of Electronics and Information Engineers
• /
• v.51 no.6
• /
• pp.221-227
• /
• 2014

In this paper, we propose a synthesis method that can effectively suppress the ambience which affects tonal components in the PS decoder. Ambience component was obtained by using decorrelation filter and the weighting of the ambience in the decoder was determined through IC parameter. However, since the parameters are extracted in the sub-band domain, a low IC value could be analyzed even if the tonal component is dominant. The quality of the output signal may be degraded. To prevent this problem, the tonality was measured in the downmixed signal and the weighting of the ambience components were adjusted appropriately according to the measured tonality index. The performance of the proposed method was evaluated by simulations. Furthermore, the subjective test was performed and the results confirmed that the proposed method offers improved quality.

• Journal of IKEEE
• /
• v.23 no.4
• /
• pp.1393-1399
• /
• 2019

With the rapid growth of the Internet, millions of indiscriminate advertising SMS are sent every day because of the convenience of sending and receiving data. Although we still use methods to block spam words manually, we have been actively researching how to filter spam in a various ways as machine learning emerged. However, spam words and patterns are constantly changing to avoid being filtered, so existing machine learning mechanisms cannot detect or adapt to new words and patterns. Recently, the concept of Lifelong Learning emerged to overcome these limitations, using existing knowledge to keep learning new knowledge continuously. In this paper, we propose a method of spam filtering system using ensemble techniques of naive bayesian which is most commonly used in document classification and LLML(Lifelong Machine Learning). We validate the performance of lifelong learning by applying the model ELLA and the Naive Bayes most commonly used in existing spam filters.

• Journal of the Institute of Electronics Engineers of Korea TC
• /
• v.37 no.1
• /
• pp.76-87
• /
• 2000

Multicarrier DS-CDAM is an effective approach to realize wideband CDMA system in a multipath fading channeal. In this paper, we propose a cinvolutionally-coded overlapped multicarrier DS-CDMA system, and show the performance improvement by comparing with conventional multicarrier DS-CDMA system. We consider the case of 50% overlapping with the adjacent subband to utilize the transmission bandwidth more efficiently. In the proposed multicarrier system, each of the rate 1/M convolutionally-encoded symbols is also 1/R repetition coded and transmitted using overlapped multicarriers and we may obtain the coding gain and frequency diversity effect, simultaneously. We also analyze the possibility of reduction in total MUI by considering a raised-cosine wave-shaping filter having a roll-off factor (0< ${\beta}{\le}1$). It will be shown that the proposed system outperforms the multicarrier DS-CDMA system in $^{[3]}$.

• Spatial Information Research
• /
• v.10 no.1
• /
• pp.61-76
• /
• 2002

Mobile map services using PDA are prevailing because of the rapid developments of techniques of the internet and handhold devices recently. While the volume of spatial data is tremendous and the spatial operations are time-intensive, the PDA has small size memory and a low performance processor. Therefore, the spatial index for PDA should be small size and efficiently filter out the candidate objects of spatial operation as well. This paper proposes a spatial index far PDA called MHF(Multilevel Hashing File). The MHF has simple structure for storage efficiency and uses a hashing technique, which is direct search method, for search efficiency. This paper also designs a compression technique for MBR. which occupies almost 80% of index data in the two dimensional case. We call it HMBR. Although the HMBR technique reduces the MB\ulcorner size to almost a third, it shows good filtering efficiency because of no information loss by quantization in case of small objects that occupy a major portion. Our experimental tests show that the proposed MHF index using HMBR technique is appropriate for PDA in terms of the size of index, the Number of MBR comparisons, the filtering efficiency and the execution time of spatial operations.

• Journal of Broadcast Engineering
• /
• v.18 no.4
• /
• pp.619-630
• /
• 2013

In this paper, we propose an edge directed color demosaicing algorithm considering color channel correlation. The proposed method consists of local region classification step and edge directional interpolation step. In the first step, each region of a given Bayer image is classified as normal edge, pattern edge, and flat regions by using intra channel and inter channel gradients. Especially, two criteria and verification process for the normal edge and pattern edge classification are used to reduce edge direction estimation error, respectively. In the second step, edge directional interpolation process is performed according to characteristics of the classified regions. For horizontal and vertical directional interpolations, missing color components are obtained from interpolation equations based on intra channel and inter channel correlations in order to improve the performance of the directional interpolations. The simulation results show that the proposed algorithm outperforms conventional approaches in both objective and subjective terms.

• Proceedings of the Korean Society of Precision Engineering Conference
• /
• 2001.04a
• /
• pp.135-139
• /
• 2001

This paper describes a string resonator that is used for the interrogation system of a Fiber Bragg Grating(FBG) strain sensor. The strain on the fiber piece is calculated from the measured frequency based on that the natural frequency of a string is a function of the applied absolute strain. Existing research considered a fiber as a string, but a fiber is not a string in the strict sense due to its bending stiffness, thus the fiber should be modeled as a beam accompanied with an axial force. In the vibration modeling, the relationship between the strain and the natural frequency is derived, and then the resonance condition is described in terms of both the phase and the mode shape for sustaining resonant motion. Several experiments verify the effectiveness of the proposed model of the fiber. The performance of the string resonator is analyzed by measuring the frequency change according to the applied strains in the dynamic range of 1100$\mu\varepsilon$ referred to the displacement from capacitance sensor. From the experimental results, the implemented string resonator provides the accuracy of $\pm$3$\mu\varepsilon$, the quasi-static resolution of ~0.1$\mu\varepsilon$(rms) which amount to be $\pm$0.17$\mu\textrm{m}$ and ~6nm respectively, in case of fiber length of 56mm. For a dynamic strain, it can provide the accuracy of ~3$\mu\varepsilon$ until the frequency comes to 8Hz. As a consequence, the string resonator proposed for FBG sensor provides the high accuracy and the high resolution in strain measurement, and also it is expecting to be used, for the application, to not only strain but also displacement measuring device.

• Journal of the Korean Society for Precision Engineering
• /
• v.30 no.9
• /
• pp.901-908
• /
• 2013

The introduction of industrial intelligent robot using vision sensor has been interested in automated factory. 2D and 3D vision sensors have used to recognize object and to estimate object pose, which is for packaging parts onto a complete whole. But it is not trivial task due to illumination and various types of objects. Object image has distorted due to illumination that has caused low reliability in recognition. In this paper, recognition method of complex shape object has been proposed. An accurate object region has detected from combined binary image, which has achieved using DoG filter and local adaptive binarization. The object has recognized using neural network, which is trained with sub-divided object class according to object type and rotation angle. Predefined shape model of object and maximal slope have used to estimate the pose of object. The performance has evaluated on ETRI database and recognition rate of 96% has obtained.

• The Journal of the Acoustical Society of Korea
• /
• v.22 no.4
• /
• pp.261-268
• /
• 2003

Conventional acoustic echo cancellers using ES algorithm have simple structure and fast convergence speed compared with those using NLMS algorithm, but they are very weak to external noise because ES algorithm updates the adaptive filter taps based on average energy reduction rate of room impulse response in specific acoustical condition. To solve this problem, in this paper, a new update algorithm for acoustic echo canceller with stepsize matrix generator is proposed. A set of stepsizes is determined based on residual error energy which is estimated by two moving average operators, and applied to the echo canceller in matrix from, resulting in improved convergence speed. Simulations in various noise condition show that the proposed algorithm improves the robustness of acoustic echo canceller to external noise.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.54 no.3
• /
• pp.70-77
• /
• 2017

Recently, Internet of Things (IoT) and deep learning techniques have affected video surveillance systems in various ways. The surveillance features that perform detection, tracking, and classification of specific objects in Closed Circuit Television (CCTV) video are becoming more intelligent. This paper presents real-time algorithm that can run in a PC environment using only a low power CPU. Traditional tracking algorithms combine background modeling using the Gaussian Mixture Model (GMM), Hungarian algorithm, and a Kalman filter; they have relatively low complexity but high detection errors. To supplement this, deep learning technology was used, which can be trained from a large amounts of data. In particular, an SRGB(Sequential RGB)-3 Layer CNN was used on tracked objects to emphasize the features of moving people. Performance evaluation comparing the proposed algorithm with existing ones using HOG and SVM showed move-in and move-out error rate reductions by 7.6 % and 9.0 %, respectively.

• Spatial Information Research
• /
• v.18 no.1
• /
• pp.77-87
• /
• 2010

Spatio-temporal join operators are essential to the management of spatio-temporal data such as moving objects. For example, the join operators are parts of processing to analyze movement of objects and search similar patterns of moving objects. Various studies on spatio-temporal join queries in outdoor space have been done. Recently with advance of indoor positioning techniques, location based services are required in indoor space as well as outdoor space. Nevertheless there is no one about processing of spatio-temporal join query in indoor space. In this paper, we introduce continuous spatio-temporal self-join queries in indoor space and propose a method of processing of the join queries over stream data of moving objects. The continuous spatio-temporal self-join query is to update the joined result set satisfying spatio-temporal predicates continuously. We assume that positions of moving objects are represented by symbols such as a room or corridor. This paper proposes a data structure, called Candidate Pairs Buffer, to filter and maintain massive stream data efficiently and we also investigate performance of proposed method in experimental study.