• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.34 no.4C
• /
• pp.388-396
• /
• 2009

This paper proposes a new fire-smoke detection method by using extracted features from camera images and pattern recognition technique. First, moving regions are detected by analyzing the frame difference between two consecutive images and generate candidate smoke regions by applying smoke color model. A smoke region generally has a few characteristics such as similar color, simple texture and upward motion. From these characteristics, we extract brightness, wavelet high frequency and motion vector as features. Also probability density functions of three features are generated using training data. Probabilistic models of smoke region are then applied to observation nodes of our proposed Dynamic Bayesian Networks (DBN) for considering time continuity. The proposed algorithm was successfully applied to various fire-smoke tasks not only forest smokes but also real-world smokes and showed better detection performance than previous method.

• Biomolecules & Therapeutics
• /
• v.8 no.2
• /
• pp.160-166
• /
• 2000

The present study examined influence of various ischemic duration on extent of focal ischemic brain injury induced by middle cerebral artery occlusion (MCAO) in rats. The MCAO was produced by insertion of a 17 mm silicone-coated 4-0 nylon surgical thread to the origin of MCA through the internal carotid artery for 30, 60, 90, 120 min (transient) or 24 hr (permanent) in male Sprague-Dawley rats under isoflurane anesthesia. Reperfusion in transient MCAO models was achieved by pulling the thread out of the internal carotid artery. Only rats showing neurological deficits characterized by left hemiparesis and/or circling to the left, were included in cerebral ischemic groups. The rats were sacrificed 24 hr after MCAO and seven serial coronal slices of the brain were stained with 2,3,5-triphenyltetrazolium chloride. Infarct size was measured using a computerized image analyzer. Ischemic damage was common in the frontoparietal cortex (somatosensory area) and the lateral segment of the striatum while damage to the medial segment of the striatum depended on the duration of the occlusion. In the 30-min MCAO grouts, however, infarcted region was primarily confined to the striatum and it was difficult to clearly delineate the region since there was mixed population of live and dead cells in the nucleus. Infarct volume was generally increased depending on the duration of MCAO, showing the most severe damage in the permanent MCAO group. However, there was no significant difference in infarct size between the 90-min and 120-min MCAO groups. % Edema also tended to increase depending on the duration of MCAO. The results suggest that the various focal ischemic rat models established in the present study can be used to evaluate in vivo neuroprotective activities of candidate compounds or to elucidate pathophysiological mechanisms of ischemic neuronal cell death.

• Journal of KIISE:Software and Applications
• /
• v.32 no.12
• /
• pp.1247-1259
• /
• 2005

This paper proposes a real-time face recognition system which detects multiple faces from low resolution video such as web-camera video. Face recognition system consists of the face detection step and the face classification step. At First, it finds face region candidates by using AdaBoost based object detection method which have fast speed and robust performance. It generates reduced feature vector for each face region candidate by using principle component analysis. At Second, Face classification used Principle Component Analysis and multi-SVM. Experimental result shows that the proposed method achieves real-time face detection and face recognition from low resolution video. Additionally, We implement the auto-tracking face recognition system using the Pan-Tilt Web-camera and radio On/Off digital door-lock system with face recognition system.

• Journal of Korean Society for Geospatial Information Science
• /
• v.16 no.1
• /
• pp.49-55
• /
• 2008

The face extraction is very important to provide the images of the roads and road sides without the problem of privacy. For face extraction form roadside images, we detected the skin color area by using HSI and YCrCb color models. Efficient skin color detection was achieved by using these two models. We used a connectivity and intensity difference for grouping, skin color regions further we applied shape conditions (rate, area, number and oval condition) and determined face candidate regions. We applied thresholds to region, and determined the region as the face if black part was over 5% of the whole regions. As the result of the experiment 28 faces has been extracted among 38 faces had problem of privacy. The reasons which the face was not extracted were the effect of shadow of the face, and the background objects. Also objects with the color similar to the face were falsely extracted. For improvement, we need to adjust the threshold.

• KIPS Transactions on Software and Data Engineering
• /
• v.7 no.2
• /
• pp.69-76
• /
• 2018

Referring expressions are natural language constructions used to identify particular objects within a scene. In this paper, we propose a new deep neural network model for referring expression comprehension. The proposed model finds out the region of the referred object in the given image by making use of the rich information about the referred object itself, the context object, and the relationship with the context object mentioned in the referring expression. In the proposed model, the object matching score and the relationship matching score are combined to compute the fitness score of each candidate region according to the structure of the referring expression sentence. Therefore, the proposed model consists of four different sub-networks: Language Representation Network(LRN), Object Matching Network (OMN), Relationship Matching Network(RMN), and Weighted Composition Network(WCN). We demonstrate that our model achieves state-of-the-art results for comprehension on three referring expression datasets.

• Journal of the Korea Society of Computer and Information
• /
• v.13 no.6
• /
• pp.77-86
• /
• 2008

The Web information sources can be regarded as the largest distributed database to the users. By virtually integrating the distributed information sources and regarding them as a single huge database, we can query the database to extract information. This capability is important to develop Web application programs. We have to infer a database schema from browsing-oriented Web documents in order to integrate databases. This paper presents a heuristic algorithm to infer the XML Schema fully automatically from semi-structured Web documents. The algorithm first extracts candidate pattern regions based on predefined structure-making tags, and determines a target pattern region using a few heuristic factors, and then derives XML Schema extraction rules from the target pattern region. The schema extraction rule is represented in XQuery, which makes development of various application systems possible using open standard XML tools. We also present the experimental results for several public web sources to show the effectiveness of the algorithm.

• Proceedings of the Korean Institute of Electrical and Electronic Material Engineers Conference
• /
• 2010.03a
• /
• pp.7-7
• /
• 2010

Energy harvesting from the environment has been of great interest as a standalone power source of wireless sensor nodes for Ubiquitous Sensor Networks(USN). In particular, the piezoelectric energy harvesting from ambient vibration sources has intensively researched because it has a relatively high power density comparing with other energy scavenging methods. Through recent advances in low power consumption RF transmitters and sensors, it is possible to adopt a micro-power energy harvesting system realized by MEMS technology for the system-on-chip. However, the MEMS energy harvesting system has some drawbacks such as a high natural frequency over 300 Hz and a small power generation due to a small dimension. To overcome these limitations, we devised a novel power generator with a spiral spring structure as shown in the figure. The natural frequency of a cantilever could be decreased to the usable frequency region (under 300 Hz) because the natural frequency depends on the length of a cantilever. In this study, the natural frequency of the energy harvester was a lower than a normal cantilever structure and sufficiently controllable in 50 - 200 Hz frequency region as adjusting weight of a proof mass. Moreover, the MEMS energy harvester had a high energy conversion efficiency using a shear mode ($d_{15}$) is much larger than a 33 mode ($d_{33}$) and the energy conversion efficiency is proportional to the piezoelectric constant (d). We expect the spiral type MEMS power generator would be a good candidate for a standalone power generator for USN.

• Journal of Korea Foundry Society
• /
• v.33 no.4
• /
• pp.157-161
• /
• 2013

A ribbon-type polycrystalline silicon wafer was directly fabricated from liquid silicon via a novel technique for both a fast growth rate and large grain size by exploiting gas pressure. Effects of processing parameters such as moving speed of a dummy bar and the length of the solidification zone on continuous casting of the silicon wafer were investigated. Silicon melt extruded from the growth region in the case of a solidification zone with a length of 1cm due to incomplete solidification. In case of a solidification zone wieh a length of 2 cm, on the other hand, continuous casting of the wafer was impossible due to the volume expansion of silicon derived from the liquid-solid transformation in solidification zone. Consequently, the optimal length of the solidification zone was 1.5 cm for maintaining the position of the solid-liquid interface in the solidification zone. The silicon wafer could be continuously casted when the moving speed of the dummy bar was 6 cm/min, but liquid silicon extruded from the growth region without solidification when the moving speed of the dummy bar was ${\geq}$ 9 cm/min. This was due to a shift of the position of the solid-liquid interface from the solidification zone to the moving area. The present study reports experimental findings on a new direct growth system for obtaining silicon wafers with both high quality and productivity, as a candidate for an alternate route for the fabrication of ribbon-type silicon wafers.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.8 no.9
• /
• pp.1319-1324
• /
• 2013

The mean shift algorithm has achieved considerable success in object tracking due to its simplicity and robustness. It finds local minima of a similarity measure between the color histograms or kernel density estimates of the target and candidate image. However, it is sensitive to the noises due to objects or background having similar color distributions. In addition, occlusion by another object often causes a face region to change in size and position although a face region is a critical clue to perform face recognition or compute face orientation. We assume that depth and color are effective to separate a face from a background and a face from objects, respectively. From the assumption we devised a bilateral filter using color and depth and incorporate it into the mean-shift algorithm. We demonstrated the proposed method by some experiments.

• The Transactions of the Korean Institute of Power Electronics
• /
• v.16 no.4
• /
• pp.358-365
• /
• 2011

Recently, the LED(Light Emitting Diode) replacing incandescent light bulbs and fluorescent light has great attentions as a most promising candidate for the next generation lighting source due to its environment-friendly characteristics, long life and excellent efficiency. Moreover, since it is a semiconductor device which can convert the electric energy to visible light at a very high speed, it can also used as a communication device. Therefore, the VLC(Visible Light Communication) using the LED can perform the near field communication and lighting function at the same time without additional expenses. However, since the switching device of the conventional LED driver for VLC is operated in the linear region, there exist several drawbacks such as a poor power conversion efficiency and serious heat generation. On the other hand, since the proposed driver is operated in the on/off switching region, it features a higher efficiency and more improved heat generation. To verify the validity of the proposed LED driver, experimental results from a prototype of 20W rated LED driver applied to 3MHz bps broadcasting audio system are given.