• The Journal of the Acoustical Society of Korea
• /
• v.29 no.8
• /
• pp.509-515
• /
• 2010

This paper presents a new method for upmixing mono signal to stereo signal with guaranteeing high stereophonic image quality (SIQ) and large apparent source width (ASW). The proposed method consists of analysis phase and synthesis phase. In analysis phase, a mono signal is first decomposed into multiple sound sources by the use of high-rank nonnegative matrix factorization. Then the multiple sources are clustered into two groups based on tonality criterion. In synthesis phase, one group is directly fed into left and right channels while the other group is decorrelated before being fed into each channel. Subjective tests reveals that the proposed method gives listener high SIQ and large ASW with minimizing timbral distortions.

• Journal of IKEEE
• /
• v.25 no.3
• /
• pp.493-500
• /
• 2021

In this paper, we propose the development of deep learning structure to improve quality of polygonal containers. The deep learning structure consists of a convolution layer, a bottleneck layer, a fully connect layer, and a softmax layer. The convolution layer is a layer that obtains a feature image by performing a convolution 3x3 operation on the input image or the feature image of the previous layer with several feature filters. The bottleneck layer selects only the optimal features among the features on the feature image extracted through the convolution layer, reduces the channel to a convolution 1x1 ReLU, and performs a convolution 3x3 ReLU. The global average pooling operation performed after going through the bottleneck layer reduces the size of the feature image by selecting only the optimal features among the features of the feature image extracted through the convolution layer. The fully connect layer outputs the output data through 6 fully connect layers. The softmax layer multiplies and multiplies the value between the value of the input layer node and the target node to be calculated, and converts it into a value between 0 and 1 through an activation function. After the learning is completed, the recognition process classifies non-circular glass bottles by performing image acquisition using a camera, measuring position detection, and non-circular glass bottle classification using deep learning as in the learning process. In order to evaluate the performance of the deep learning structure to improve quality of polygonal containers, as a result of an experiment at an authorized testing institute, it was calculated to be at the same level as the world's highest level with 99% good/defective discrimination accuracy. Inspection time averaged 1.7 seconds, which was calculated within the operating time standards of production processes using non-circular machine vision systems. Therefore, the effectiveness of the performance of the deep learning structure to improve quality of polygonal containers proposed in this paper was proven.

• Journal of IKEEE
• /
• v.26 no.3
• /
• pp.408-415
• /
• 2022

In this paper, we propose a 3D point cloud reconstruction technique from 2D images using efficient feature map extraction network. The originality of the method proposed in this paper is as follows. First, we use a new feature map extraction network that is about 27% efficient than existing techniques in terms of memory. The proposed network does not reduce the size to the middle of the deep learning network, so important information required for 3D point cloud reconstruction is not lost. We solved the memory increase problem caused by the non-reduced image size by reducing the number of channels and by efficiently configuring the deep learning network to be shallow. Second, by preserving the high-resolution features of the 2D image, the accuracy can be further improved than that of the conventional technique. The feature map extracted from the non-reduced image contains more detailed information than the existing method, which can further improve the reconstruction accuracy of the 3D point cloud. Third, we use a divergence loss that does not require shooting information. The fact that not only the 2D image but also the shooting angle is required for learning, the dataset must contain detailed information and it is a disadvantage that makes it difficult to construct the dataset. In this paper, the accuracy of the reconstruction of the 3D point cloud can be increased by increasing the diversity of information through randomness without additional shooting information. In order to objectively evaluate the performance of the proposed method, using the ShapeNet dataset and using the same method as in the comparative papers, the CD value of the method proposed in this paper is 5.87, the EMD value is 5.81, and the FLOPs value is 2.9G. It was calculated. On the other hand, the lower the CD and EMD values, the better the accuracy of the reconstructed 3D point cloud approaches the original. In addition, the lower the number of FLOPs, the less memory is required for the deep learning network. Therefore, the CD, EMD, and FLOPs performance evaluation results of the proposed method showed about 27% improvement in memory and 6.3% in terms of accuracy compared to the methods in other papers, demonstrating objective performance.

• Journal of the Korean Institute of Telematics and Electronics D
• /
• v.36D no.2
• /
• pp.20-30
• /
• 1999

We have made the single side hand filter for the dual channel receiver which is a heterodyne receiver to observe the cosmic radio waves with 100GHz band ranged from 85GHz to 115GHz and 150GHz band ranged from 125GHz to 175GHz simultaneously. We have introduced the filter theory using the principle of the Martion-Puplett interferometer, which has the characteristics of rotated polarization. To reduce the loss of the transmission and beam coupling which are caused from the path difference associated with the intermediate frequency the design and the implementation have been intensely considered. The receiver needs two filters with different characteristics each other. Because each of them has the optimum positions as a function frequency at which the signal frequency is fed to mixer and the image frequency is rejected to the image termination load. The intermediate frequency and its band width have been also evaluated. We have measured the property of two filters using the vector network analyser and the beam measurement system which is made by us. The responses of the filter as a function of the position and the frequency are compared with the theory. It is shown that not only the measured values are very close to the theoretical values, but also the image rejection ratios are better than 22dB for both filters. Through successful observation using a dual channel receiver with two manufactured filters, the performance of the filters has finally verified.

• Geophysics and Geophysical Exploration
• /
• v.21 no.1
• /
• pp.26-32
• /
• 2018

The Northeast China is an important site geologically and geophysically because of a huge volcano called Mt. Baekdu, which is one of the largest volcanoes in the world. Signs of eruption have been recently observed and people are keen to its behavior. We carried out relative travel time tomography to investigate the velocity structure between 100 ~ 600 km depth beneath Northeast China. We used teleseismic data during 2009 ~ 2011 recorded in NecessArray provided by IRIS (Incorporated Research Institute for Seismology). The relative observations were obtained by using the multi-channel cross-correlation method. Based on the tomographic results, we observed that the locations beneath which low-velocity zones are observed coincide with the locations of several volcanic regions in Northeast China. A low-velocity anomaly is revealed beneath Mt. Baekdu down to 600 km depth, which is thought to the main origin of the magma supply for Mt. Baekdu. Another low velocity anomaly is observed beneath east of the Datong volcano down to around 300 km depth, which is inferred to be related to an upwelling from deep mantle. We observed a low velocity anomaly beneath the Wudalianchi volcano down to around 200 km depth, which may imply that this volcano has been formed by an upwelling from the asthenosphere.

• Journal of the Institute of Convergence Signal Processing
• /
• v.19 no.1
• /
• pp.7-13
• /
• 2018

In the current study, we proposed an optimized brain-computer interface (BCI) which employed blind source separation (BBS) approach to remove noises. Thus motor imagery (MI) signal and steady state visual evoked potential (SSVEP) signal were easily to be detected due to enhancement in signal-to-noise ratio (SNR). Moreover, a combination between MI and SSVEP which is typically can increase the number of commands being generated in the current BCI. To reduce the computational time as well as to bring the BCI closer to real-world applications, the current system utilizes a single-channel EEG signal. In addition, a convolutional neural network (CNN) was used as the multi-class classification model. We evaluated the performance in term of accuracy between a non-BBS+BCI and BBS+BCI. Results show that the accuracy of the BBS+BCI is achieved $16.15{\pm}5.12%$ higher than that in the non-BBS+BCI by using BBS than non-used on. Overall, the proposed BCI system demonstrate a feasibility to be applied for multi-dimensional control applications with a comparable accuracy.

• Geophysics and Geophysical Exploration
• /
• v.22 no.4
• /
• pp.225-238
• /
• 2019

Migration velocity analysis (MVA) for creating optimum depth-domain velocities in seismic imaging was applied to marine long-offset multi-channel data, and the effectiveness of the MVA approach was demonstrated by the combinations of conventional data processing procedures. The time-domain images generated by conventional time-processing scheme has been considered to be sufficient so far for the seismic stratigraphic interpretation. However, when the purpose of the seismic imaging moves to the hydrocarbon exploration, especially in the geologic modeling of the oil and gas play or lead area, drilling prognosis, in-place hydrocarbon volume estimation, the seismic images should be converted into depth domain or depth processing should be applied in the processing phase. CMP-based velocity analysis, which is mainly based on several approximations in the data domain, inherently contains errors and thus has high uncertainties. On the other hand, the MVA provides efficient and somewhat real-scale (in depth) images even if there are no logging data available. In this study, marine long-offset multi-channel seismic data were optimally processed in time domain to establish the most qualified dataset for the usage of the iterative MVA. Then, the depth-domain velocity profile was updated several times and the final velocity-in-depth was used for generating depth images (CRP gather and stack) and compared with the images obtained from the velocity-in-time. From the results, we were able to confirm the depth-domain results are more reasonable than the time-domain results. The spurious local minima, which can be occurred during the implementation of full waveform inversion, can be reduced when the result of MVA is used as an initial velocity model.

• Journal of Intelligence and Information Systems
• /
• v.28 no.3
• /
• pp.119-137
• /
• 2022

Recent recommendation system studies apply various deep learning models to represent user and item interactions better. One of the noteworthy studies is ONCF(Outer product-based Neural Collaborative Filtering) which builds a two-dimensional interaction map via outer product and employs CNN (Convolutional Neural Networks) to learn high-order correlations from the map. However, ONCF has limitations in recommendation performance due to the problems with CNN and the absence of side information. ONCF using CNN has an inductive bias problem that causes poor performances for data with a distribution that does not appear in the training data. This paper proposes to employ a Vision Transformer (ViT) instead of the vanilla CNN used in ONCF. The reason is that ViT showed better results than state-of-the-art CNN in many image classification cases. In addition, we propose a new architecture to reflect side information that ONCF did not consider. Unlike previous studies that reflect side information in a neural network using simple input combination methods, this study uses an independent auxiliary classifier to reflect side information more effectively in the recommender system. ONCF used a single latent vector for user and item, but in this study, a channel is constructed using multiple vectors to enable the model to learn more diverse expressions and to obtain an ensemble effect. The experiments showed our deep learning model improved performance in recommendation compared to ONCF.

• The KIPS Transactions:PartB
• /
• v.18B no.6
• /
• pp.365-374
• /
• 2011

Recently, malicious video classification and filtering techniques are of practical interest as ones can easily access to malicious multimedia contents through the Internet, IPTV, online social network, and etc. Considerable research efforts have been made to developing malicious video classification and filtering systems. However, the malicious video classification and filtering is not still being from mature in terms of reliable classification/filtering performance. In particular, the most of conventional approaches have been limited to using only the spatial features (such as a ratio of skin regions and bag of visual words) for the purpose of malicious image classification. Hence, previous approaches have been restricted to achieving acceptable classification and filtering performance. In order to overcome the aforementioned limitation, we propose new malicious video classification framework that takes advantage of using both the spatial and temporal features that are readily extracted from a sequence of video frames. In particular, we develop the effective temporal features based on the motion periodicity feature and temporal correlation. In addition, to exploit the best data fusion approach aiming to combine the spatial and temporal features, the representative data fusion approaches are applied to the proposed framework. To demonstrate the effectiveness of our method, we collect 200 sexual intercourse videos and 200 non-sexual intercourse videos. Experimental results show that the proposed method increases 3.75% (from 92.25% to 96%) for classification of sexual intercourse video in terms of accuracy. Further, based on our experimental results, feature-level fusion approach (for fusing spatial and temporal features) is found to achieve the best classification accuracy.

• Journal of the Korean Institute of Illuminating and Electrical Installation Engineers
• /
• v.21 no.7
• /
• pp.37-45
• /
• 2007

Recently, there has been an increase of the use of ground system for lightning protection called deeply driven grounding electrode. In the case of deeply driven grounding electrode, the rod electrode is equipped perpendicularly and deeply, therefore, it has a benefit to have less restriction of place compared to mesh grid electrode. However, ground impedance is largely changed by the local earth resistivity, so it requires a detailed analysis of the ground structure when planning. The measurement of earth resistivity by existing Wenner's method has been widely used, however, this method can not find out a change in the local ground resistance and it shows the result outwardly to be difficult to estimate exact depth. Therefore, this study analyzed the ground structure as 2-D image using 96 channels measurement facility and tried to analyze change in the local ground resistance and depth of the ground in order to design a deeply driven electrode effectively for lightning protection. It used Wenner alpha method dipole-dipole method and Schlumberger method for 2-D image analysis of the ground resistivity ma based on, it the result was compared with the ground structure analyzed with the result using the CDEGS and Wenner 1-D method.