• Journal of Broadcast Engineering
• /
• v.25 no.2
• /
• pp.200-207
• /
• 2020

Recently, deep neural networks such as CNN are showing excellent performance in various fields such as image classification, object recognition, visual quality enhancement, etc. However, as the model size and computational complexity of deep learning models for most applications increases, it is hard to apply neural networks to IoT and mobile environments. Therefore, neural network compression algorithms for reducing the model size while keeping the performance have been being studied. In this paper, we apply few compression methods to CNN models and evaluate their performances in the embedded environment. For evaluate the performance, the classification performance and inference time of the original CNN models and the compressed CNN models on the image inputted by the camera are evaluated in the embedded board equipped with QCS605, which is a customized AI chip. In this paper, a few CNN models of MobileNetV2, ResNet50, and VGG-16 are compressed by applying the methods of pruning and matrix decomposition. The experimental results show that the compressed models give not only the model size reduction of 1.3~11.2 times at a classification performance loss of less than 2% compared to the original model, but also the inference time reduction of 1.2~2.21 times, and the memory reduction of 1.2~3.8 times in the embedded board.

• Journal of Fashion Business
• /
• v.7 no.1
• /
• pp.116-134
• /
• 2003

The beauty industry of today tends to gradually develope as one of the important marketing strategies along with the total fashion sector. Hair color, being one of the factors of hair style, especially builds his or her own image and even becomes the nonverbal communication method which let others know him/herself. Therefore, women's needs and expectations of hair coloring are varied. To meet these need, it is urgent to figure out, most of all, the status of customers' awareness about hair coloring. Thus, in this study, research and analysis on hair coloring were made, focusing on the women utilizing 20 beauty shops in Changwon city. And the goal that this study set is to use as the study material for the hair coloring consulting in the beauty industry and the basic data for marketing in the industry job-sites. The conclusion by the three factors(age, job, monthly income) were as follows: 1. Dandyism was on the 1st order in the reason for coloring hair regardless of age, job, and monthly income. This seems to be speaking for the change of women's awareness of coloring hair. 2. The subject of making decision upon whether she was going to color her hair or not was herself regardless of age, job, and monthly income. Among the people belong to these three factors, brown was the color in overall preference. 3. More than half of those who belong to the three factors were not aware of the hair color, but their senses and interest in the color in vogue were high. 4. In terms of the hair coloring periodic time, many people had $2{\sim}3$months, but it was possible to know that they were interested in improving their images, considering the fact that most of the salaried people had their hair dyed within a month and the people in service job over a month. 5. The acquaintance around me had the strongest effect in collecting information from the people in those three factors and mass media was the next. The results of this study, in my opinion, would help the customers have beautiful and healthy hair and make images of their own by making the hair designers develope professional knowledges and skills on the hair coloring and improve the quality of beauty services.

• Journal of Korea Multimedia Society
• /
• v.7 no.2
• /
• pp.251-263
• /
• 2004

The Synchronized Multimedia Integration Language (SMIL), recommended by the World Wide Web Consortium (W3C) in 1998, is an XML-based declarative language to synchronize and present multimedia documents. SMIL can create new multimedia data integrating various types of multimedia objects which exist separately such as text, video, graphics and audio. It can support synchronization of multimedia data which are limited in current HTML-based Web technology. For its popularity, it is required to develop a multimedia server guaranteeing Quality of Service (QoS), authoring tool and player. For developing a SMIL authoring tool and player, the technologies are essentially required to read and analyze a SMIL document and to play synchronized various types of media objects in a timeline. In this paper, we describe a design and implementation of an event processor which supports SMIL 2.0 timing model. Moreover, we also develop a SMIL 2.0 player using the proposed event processor. This will facilitate the play of SMIL contents, so that it can contribute to the prosperity of SMIL technology It is possible to reuse in various language profiles defined in the SMIL standard. This player is expected to be utilized in other standard integrating SMIL such as XHTML＋SMIL and SMIL Animation.

• Journal of Broadcast Engineering
• /
• v.12 no.1 s.34
• /
• pp.11-27
• /
• 2007

T-DMB(Terrestrial-Digital Multimedia Broadcasting) for portable and mobile broadcasting service was officially launched in the metropolitan area of Korea in December 2005. The development of interactive data services and the expansion of T-DMB network have been now progressed, and the regional T-DMB broadcasters will be selected in the near future. Although it is important to evaluate service area and reduce a fringe area for optimizing RF coverage of T-DMB, there was no professional system to support those works, and therefore lots of resources in budget, manpower and time were required. The measurement considering characteristics of SFN(Single Frequency Network) is essential in the implementation of T-DMB network which needs to control synchronization, spacing and output power of transmitters. In addition, mobile measurement, enough measured parameters for many-sided analysis of reception quality, efficient management of enormous measured data and representation of measured results on the electronic map are also important in evaluating service area. In this paper, we derived requirements for a new measurement system in T-DMB by considering the above details, and we proposed and developed an integrated measurement and analysis system. The developed system was applied to the implementation of T-DMB network of KBS and field tests, and it proved its efficiency and accuracy in result.

• Journal of Broadcast Engineering
• /
• v.21 no.6
• /
• pp.957-964
• /
• 2016

Acquisition of reliable depth maps is a critical requirement in many applications such as 3D videos and free-viewpoint TV. Depth information can be obtained from the object directly using physical sensors, such as infrared ray (IR) sensors. Recently, Time-of-Flight (ToF) range camera including KINECT depth camera became popular alternatives for dense depth sensing. Although ToF cameras can capture depth information for object in real time, but are noisy and subject to low resolutions. Recently, filter-based depth up-sampling algorithms such as joint bilateral upsampling (JBU) and noise-aware filter for depth up-sampling (NAFDU) have been proposed to get high quality depth information. However, these methods often lead to texture copying in the upsampled depth map. To overcome this limitation, we formulate a convex optimization problem using higher order regularization for depth map upsampling. We decrease the texture copying problem of the upsampled depth map by using edge weighting term that chosen by the edge information. Experimental results have shown that our scheme produced more reliable depth maps compared with previous methods.

• Journal of Broadcast Engineering
• /
• v.23 no.3
• /
• pp.421-430
• /
• 2018

Many voice recognition systems use methods such as MFCC, HMM to acknowledge human voice. This recognition method is designed to analyze only a targeted sound which normally appears between a human and a device one. However, the recognition capability is limited when there is a group sound formed with diversity in wider frequency range such as dog barking and indoor sounds. The frequency of overlapped sound resides in a wide range, up to 20KHz, which is higher than a voice. This paper proposes the new recognition method which provides wider frequency range by conjugating the Wideband Sound Spectrogram and the Keras Sequential Model based on DNN. The wideband sound spectrogram is adopted to analyze and verify diverse sounds from wide frequency range as it is designed to extract features and also classify as explained. The KSM is employed for the pattern recognition using extracted features from the WSS to improve sound recognition quality. The experiment verified that the proposed WSS and KSM excellently classified the targeted sound among noisy environment; overlapped sounds such as dog barking and indoor sounds. Furthermore, the paper shows a stage by stage analyzation and comparison of the factors' influences on the recognition and its characteristics according to various levels of noise.

• Journal of Broadcast Engineering
• /
• v.14 no.4
• /
• pp.463-474
• /
• 2009

By employing the subscriber concept in broadcasting services, IPTV (Internet Protocol Television) operators provide various grades of services to subscribers based on the billing level of the subscribers. With the income from subscribers for a basis, IPTV operators plan to provide high quality services. Since Web browser-based IPTV provides T-commerce and E-commerce services as well as television services, users may frequently visit other service domains to buy goods or content. To provide the user with charged or private services, these service domains request authentication of user. The existing authentication system is not appropriate for the IPTV service environment because the environment unavoidably forces the user to cross from one authentication-based service domain to another. Single sign-on provides a user with transparent authentication services by enabling an authenticated user to move between authentication-based service domains without any re-authentication. Like this distributed environment, since the IPTV service environment also provides a variety of authentication-based services, transparent authentication service needs to be provided to subscribers who want to access charged or private services. In this paper, we propose a new user authentication scheme for the IPTV environment. This scheme integrates the Security Assertion Markup Language (SAML), which is a standard for XML-based single sign on. We validate this scheme using a simple use case scenario.

• Microbiology and Biotechnology Letters
• /
• v.39 no.4
• /
• pp.357-363
• /
• 2011

This study focus was primarily the development of liquid starters for Aspergillus oryzae and Aspergillus niger prepared with wheat bran as a low cost culture medium. For the preparation of the liquid media wheat bran was added at rates of 0, 5, 10, 15 and 20% and the Aspergillus sp. strains were then inoculated to these prepared broths. The results indicated that the more that wheat bran was contained in the medium, the more mycelia was produced for A. oryzae and A. niger. The highest enzyme activities were obtained with a 10~15% adding rate of wheat bran for both strains. Changes in the enzyme activities of the liquid starters during various incubation times (0, 24, 48, 72 and 96 hrs), indicated that the highest enzyme activities were seen between 48 and 72 hrs of culture. In addition, a comparative study was carried out on the production of enzymes using wheat as a substrate in nuruk, with liquid starter made from fermented agents according to the same concentrations used with the wheat bran. The pH, acidity, amino acidity, reducing sugar content and enzyme activity (${\alpha}$-amylase, glucoamylase, acidic protease) of wheat nuruk made with liquid starter were compared with those of wheat nuruk made with solid starter. The results suggest that the liquid starter is superior in both cases.

• Archives of design research
• /
• v.20 no.1 s.69
• /
• pp.229-242
• /
• 2007

In Japan, a country renowned for her traditional respect for handicraft, superior craft techniques have been transmitted from generation to generation. Modern Japanese textiles reinterprete these traditional techniques in a new light and combine them with modern technology, to make modern Japanese industrial textile both futuristic and traditional simultaneously. This combining of new technology and handicraft serves as a reservoir for a remarkable diversity and originality in the development of new technology and design, and constitutes an important trend in modern industrial textiles. The aim of this study is to investigate the developmental background of Japanese textile industry that championed this trend to make the Japanese development unique in modern textile industries and to consider the concept of modern textiles as well as the structural innovation as represented by this new trend, thus contributing to the development of creative textiles for the present and the future. The findings of this study can be summarized as follows. The combining of new technology and traditional handicraft: 1) was made possible by Japanese consciousness that values traditional culture and by early recognition of advanced textile industry, which awarded new meaning and value to the traditional handicraft leading to an expanded social role for textiles. 2) helped construct a modern concept of textiles by pursuing textile media that are state-of-the-art, aesthetic, and highly sensitive to satisfy the requirements of modern industrial textile such as high quality, high performance, and high sensitivity. 3) enabled new design ideas and creation of new styles by implementing structural innovation in industrial textiles that involved diversification of expression, diversity in materials, and emphasis on texture. In short, by combining new technology and traditional handicraft, Japan came up with a new sense of aesthetics that is highly original and high level and with which the rest of the world can easily sympathize, thereby presenting one of the most important guiding principles for future textiles.

• The Journal of Engineering Geology
• /
• v.14 no.3 s.40
• /
• pp.301-311
• /
• 2004

This study is suggested a new dielectric tracer test method to understand geological structure of porous media and groundwater flow to use the dielectric constant which is one of electrical special quality of various geological materials. To measure their parameters, tracer material is made an ethanol mixing liquid(EML) having a same specific gravity of water. Also, soil materials are prepared a dielectric tracer test using the FDR system that could measure dielectric constant for saturated standard sand and river sand layers which have different initial porosity. To compare with their results, we discussed with the concentration variation of saline water having a saline concentration $3\%$ which is general tracer material by using the electro multi-meter system in the laboratory or field test. In two tracer experiment results, EML tracer test could confirm definitely EML concentration variation from each saturated soil layer as standard and river sands. However, tracer test of saline water $3\%$ concentration could not confirm permeating movement of water by degree of salinity change because these are settled at lower part column in a whole column area continuously. These causes are that specific gravity of saline water is heavier than water. That is, it could know that deposition of saline water is composed of lower part of soil column continuously independently of the direction of water into saturated soil material.