• MALSORI
• /
• no.63
• /
• pp.125-137
• /
• 2007

Recently, many techniques have been proposed to improve the noise robustness for speaker verification. In this paper, we consider the feature recombination technique in multi-band approach. In the conventional feature recombination for speaker verification, to compute the likelihoods of speaker models or universal background model, whole feature components are used. This computation method is not effective in a view point of multi-band approach. To deal with non-effectiveness of the conventional feature recombination technique, we introduce a subband likelihood computation, and propose a modified feature recombination using subband likelihoods. In decision step of speaker verification system in noise environments, a few very low likelihood scores of a speaker model or universal background model cause speaker verification system to make wrong decision. To overcome this problem, a reliable feature selection method is proposed. The low likelihood scores of unreliable feature are substituted by likelihood scores of the adaptive noise model. In here, this adaptive noise model is estimated by maximum a posteriori adaptation technique using noise features directly obtained from noisy test speech. The proposed method using subband-based reliable feature selection obtains better performance than conventional feature recombination system. The error reduction rate is more than 31 % compared with the feature recombination-based speaker verification system.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.17 no.3
• /
• pp.29-45
• /
• 2017

Ad hoc sensor network is characterized by decentralized structure and ad hoc deployment. Sensor networks have all basic features of ad hoc network except different degrees such as lower mobility and more stringent energy requirements. Existing protocols provide different tradeoffs among some desirable characteristics such as fault tolerance, distributed computation, robustness, scalability and reliability. wireless protocols suggested so far are very limited, generally focusing on communication to a single base station or on aggregating sensor data. The main reason having such restrictions is due to maximum lifetime to maintain network activities. The network lifetime is an important design metric in ad hoc networks. Since every node does a router role, it is not possible for other nodes to communicate with each other if some nodes do not work due to energy lack. In this paper, we suggest an experimental ad-hoc on-demand distance vector routing protocol to optimize the communication of energy of the network nodes.The load distribution avoids the choice of exhausted nodes at the route selection phase, thus balances the use of energy among nodes and maximizing the network lifetime. In transmission control phase, there is a balance between the choice of a high transmission power that lead to increase in the range of signal transmission thus reducing the number of hops and lower power levels that reduces the interference on the expense of network connectivity.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.46 no.5
• /
• pp.57-70
• /
• 2009

A CAD design drawing based on vector data that is very important art work in industrial fields has been considered to content that the copyright protection is urgently needed. This paper presents a watermarking scheme based on k-means++ for CAD design drawing. One CAD design drawing consists of several layers and each layer consists of various geometric objects such as LINE, POLYLINE, CIRCLE, ARC, 3DFACE and POLYGON. POLYLINE with LINE, 3DFACE and ARC that are fundamental objects make up the majority in CAD design drawing. Therefore, the proposed scheme selects the target object with high distribution among POLYLINE, 3DFACE and ARC objects in CAD design drawing and then selects layers that include the most target object. Then we cluster the target objects in the selected layers by using k-means++ and embed the watermark into the geometric distribution of each group. The geometric distribution is the normalized length distribution in POLYLINE object, the normalized area distribution in 3DFACE object and the angle distribution in ARC object. Experimental results verified that the proposed scheme has the robustness against file format converting, layer attack as well as various geometric editing provided in CAD editing tools.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.42 no.4 s.304
• /
• pp.33-42
• /
• 2005

This paper proposes a classification method for gene expression data, using membership function and neural network. The gene expression is a process to produce mRNA and protains which generate a living body, and the gene expression data is important to find out the functions and correlations of genes. Such gene expression data can be obtained from DNA 칩 massively and quickly. However, thousands of gene expression data may not be useful until it is well organized. Therefore a classification method is necessary to find the characteristics of gene data acquired from the gene expression. In the proposed method, a set of gene data is extracted according to the fisher's criterion, because we assume that selected gene data is the well-classified data sample. However, the selected gene data does not guarantee well-classified data sample and we calculate feature values using membership function to reduce the influence of outliers in gene data. Feature vectors estimated from the selected feature values are used to train back propagation neural network. The experimental results show that the clustering performance of the proposed method has been improved compared to other existing methods in various gene expression data.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.43 no.1 s.307
• /
• pp.27-34
• /
• 2006

In this paper, we propose a combined region-based image retrieval system using EHD(Edge Histogram Descriptor) and CLD(Color Layout Descriptor) of MPEG-7 descriptors. The combined descriptor can efficiently describe edge and color features in terms of sub-image regions. That is, the basic unit for the selection of the region-of-interest (ROI) in the image is the sub-image block of the EHD, which corresponds to 16 (i.e., $4{\times}4)$ non-overlapping image blocks in the image space. This implies that, to have a one-to-one region correspondence between ELE and CLD, we need to take an $8{\times}8$ inverse DCT (IDCT) for the CLD. Experimental results show that the proposed retrieval scheme can be used for image retrieval with the ROI based image retrieval for MPEG-7 indexed images.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.18 no.10
• /
• pp.551-558
• /
• 2017

To study and validate tissue-specific promoters and vectors, it is important to develop cell culture systems that retain the tissue and species specificity. Such systems are attractive alternatives to transgenic animal models. This study established a line of porcine mammary gland epithelial cells (PMECs) from a primary culture based on the cellular morphology and mRNA levels of porcine beta-casein (CSN2). The selected PMECs were stained with the cytokeratin antibody, and were shown to express milk protein genes (CSN2, lactoferrin, and whey acidic protein). In addition, to confirm the acini structure of PMEC932-7 in 3D culture, live cells were stained with SYTO-13 dye, which binds to nucleic acid. The acini of these PMECs on matrigel were formed by the aggregation of peripheral cells and featured a hollow lumens. The system was demonstrated by testing the effects of the culture conditions to cell culture including cell density and matrigel methods of the PMECs. These results suggest that PMECs possess the genetic and structural features of mammary epithelial cells.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.20 no.1
• /
• pp.116-122
• /
• 2010

This paper proposes a method for classifying targets robust to geometric transformations of targets such as rotation, scale change, translation, and pose change. Targets which have rotation, scale change, and shift is firstly classified based on CM(Confidence Map) which is generated by similarity, scale ratio, and range of orientation for SIFT(Scale-Invariant Feature Transform) feature vectors. On the other hand, DB(DataBase) which is acquired in various angles is used to deal with pose variation of targets. Range of the angle is determined by comparing and analyzing the execution time and performance for sampling intervals. We experiment on various images which is geometrically changed to evaluate performance of proposed target classification method. Experimental results show that the proposed algorithm has a good classification performance.

• Journal of Korea Multimedia Society
• /
• v.3 no.2
• /
• pp.121-132
• /
• 2000

Most of video indexing applications depend on fast and efficient archiving, browsing, retrieval techniques. A number of techniques have been approached about only pixel domain analysis until now. Those approaches brought about the costly overhead of decompressing because the most of multimedia data is typically stored in compressed format. But with a compressed video data, if we can analyze the compressed data directly. then we avoid the costly overhead such as in pixel domain. In this paper, we analyze the information of compressed video stream directly, and then extract the available features for video indexing. We have derived the technique for cut detection using these features, and the stream is divided into shots. Also we propose a new brief key frame selection technique and an efficient video indexing method using the spatial informations(DT coefficients) and also the temporal informations(motion vectors).

• The Journal of the Korea Contents Association
• /
• v.9 no.6
• /
• pp.1-8
• /
• 2009

There are a lot of influences, such as location of camera, luminosity, brightness, and direction of light, which affect the performance of 2-dimensional image recognition. This paper suggests an adaptive method for face-image recognition in noisy environments using evolvable filtering and feature extraction which uses one sample image from camera. This suggested method consists of two main parts. One is the environmental-adjustment module which determines optimum sets of filters, filter parameters, and dimensions of features by using "steady state genetic algorithm". The other another part is for face recognition module which performs recognition of face-image using the previous results. In the processing, we used Gabor wavelet for extracting features in the images and k-Nearest Neighbor method for the classification. For testing of the adaptive face recognition method, we tested the adaptive method in the brightness noise, in the impulse noise and in the composite noise and verified that the adaptive method protects face recognition-rate's rapidly decrease which can be occurred generally in the noisy environments.

• Journal of KIISE
• /
• v.42 no.12
• /
• pp.1535-1543
• /
• 2015

The brain-computer interface obtains a user's electroencephalogram as a replacement communication unit for the disabled such that the user is able to control machines by simply thinking instead of using hands or feet. In this paper, we propose a feature extraction method based on a non-selected filter by SBCSP to classify motor imagery EEG. First, we divide frequencies (4~40 Hz) into 4-Hz units and apply CSP to each Unit. Second, we obtain the FLD score vector by combining FLD results. Finally, the FLD score vector is projected onto the optimal plane for classification using PCA. We use BCI Competition III dataset IVa, and Extracted features are used as input for LS-SVM. The classification accuracy of the proposed method was evaluated using $10{\times}10$ fold cross-validation. For subjects 'aa', 'al', 'av', 'aw', and 'ay', results were $85.29{\pm}0.93%$, $95.43{\pm}0.57%$, $72.57{\pm}2.37%$, $91.82{\pm}1.38%$, and $93.50{\pm}0.69%$, respectively.