• Journal of The Korean Society For Urban Railway
• /
• v.6 no.4
• /
• pp.269-278
• /
• 2018

Railway obstacle detecion system has been introduced with high-speed railway in 2004 to prevent accidents by obstacles such as landslide, rockfall and things fallen from the gauntry over the railway. But existing system has some limitation for landslide or fallen obstacle over railway. Therefore, In this study, we suggest new advanced obstacle detection system introducing the OTDR, optical fiber fences and detection cameras. This system can detect depression degree by the force to the fences and video for the specific region as well as detection wire Off condition. We produce and functional tests for fiber fence and OTDR, which are the core parts of the development system, and results were obtained to demonstrate improved detection capabilities. Several functions also been tested to verify the advanced detection performance and got some satisfactory results. Further we will conduct environment tests and field test.

• Journal of Internet Computing and Services
• /
• v.20 no.5
• /
• pp.57-65
• /
• 2019

Creative works such as webtoon and web novel are part of property rights. However, illegal copies of them are distributed on the internet easily, which raises social issues in today's society. In order to tackle these problems, this paper proposes and presents a blockchain based copyright management system that ensures forgery prevention, robust security features, improving trading performance, cost-effective, and enhanced visibility. The system allows a user to register creative works formally just the same as before registration and also to register simple creative ideas just anytime. In the latter case, if an idea or a thought flashes across through somebody's mind, he or she can register it to the system immediately without formal registration process and afterward, can utilize a way to prove its originality through the system. Regarding large size images and video files of creative works, the system reduces data size and storage volume sharply to be processed by network entities by storing original creative works separately and including only the hash result of creative works to the transactions.

• Journal of Broadcast Engineering
• /
• v.25 no.6
• /
• pp.882-897
• /
• 2020

Recently, the use of immersive media contents representing Panorama and 360° video is increasing. Since the viewing angle is limited to generate the content through a general camera, image stitching is mainly used to combine images taken with multiple cameras into one image having a wide field of view. However, if the parallax between the cameras is large, parallax distortion may occur in the stitched image, which disturbs the user's content immersion, thus an image stitching overcoming parallax distortion is required. The existing Seam Optimization based image stitching method to overcome parallax distortion uses energy function or object segment information to reflect the location information of objects, but the initial seam generation location, background information, performance of the object detector, and placement of objects may limit application. Therefore, in this paper, we propose an image stitching method that can overcome the limitations of the existing method by adding a weight value set differently according to the type of object to the energy value using object detection based on deep learning.

• Journal of the Institute of Convergence Signal Processing
• /
• v.21 no.2
• /
• pp.55-60
• /
• 2020

In this paper, we propose a fast block matching algorithm of motion estimation using early detection of optimal candidate with high priority and a threshold. Even though so many fast algorithms for motion estimation have been published to reduce computational reduction full search algorithm, still so many works to improve performance of motion estimation are being reported. The proposed algorithm calculates block matching error for each candidate with high priority from previous partial matching error. The proposed algorithm can be applied additionally to most of conventional fast block matching algorithms for more speed up. By doing that, we can find the minimum error point early and get speed up by reducing unnecessary computations of impossible candidates. The proposed algorithm uses smaller computation than conventional fast full search algorithms with the same prediction quality as the full search algorithm. Experimental results shows that the proposed algorithm reduces 30~70% compared with the computation of the PDE and full search algorithms without any degradation of prediction quality and further reduces it with other fast lossy algorithms.

• Journal of Korean Dental Science
• /
• v.13 no.2
• /
• pp.43-51
• /
• 2020

Purpose: The practical examination portion of the National Dental Licensing Examination (NDLE) is slated to be administered in the latter half of 2021 in the form of a clinical performance examination that comprehensively evaluates the patient-dentist interaction using standardized patients (SPs). The SPs should be equipped with the basic qualities and capacity as evaluators for a fair and reliable administration of the test. Materials and Methods: In this study, we analyzed the existing training materials for SPs who participated in domestic and overseas practical tests for the development of training materials for SPs through seminars and surveys of 11 dentistry schools and colleges. Result: First, SPs should be selected according to the basic quality criteria and capacity, which they must possess, and the preliminary basic training about the details which they must have knowledge of and be provided through videorecorded cases before the implementation of the preliminary field training. Second, the roles of SPs and the calibration process of the evaluation result forms are needed when conducting the preliminary field training for SPs. After watching video-recorded scenario cases, the SPs participate in discussions about the watched videos before proceeding to calibration practices of evaluation result forms. Third, because the Type A questionnaire of the practical examination of the NDLE is dependent on the SPs' capacity and training, the fairness of the practical test is largely dependent on the SPs. Therefore, practicing the roles as evaluators and evaluation training should be provided using practical test items that can improve the reliability of the test and show a high level of reproducibility about the same case. Conclusion: The findings of this study will be utilized for the development of training materials for SPs, so they can participate in the administration of a fair and reliable practical examination of the NDLE.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.25 no.1
• /
• pp.146-151
• /
• 2021

Currently, the demand for vehicles from one family is increasing in many countries at home and abroad, reducing the number of people on the vehicle and increasing the number of vehicles on the road. The multi-passenger lane system, which is available to solve the problem of traffic congestion, is being implemented. The system allows police to monitor fast-moving vehicles with their own eyes to crack down on illegal vehicles, which is less accurate and accompanied by the risk of accidents. To address these problems, applying deep learning object recognition techniques using images from road sites will solve the aforementioned problems. Therefore, in this paper, we compare and analyze the performance of existing deep learning models, select a deep learning model that can identify real-time vehicle occupants through video, and propose a vehicle occupancy detection algorithm that complements the object-ident model's problems.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.30 no.6
• /
• pp.1053-1065
• /
• 2020

With the recent development of hardware performance and artificial intelligence technology, sophisticated fake videos that are difficult to distinguish with the human's eye are increasing. Face synthesis technology using artificial intelligence is called Deepfake, and anyone with a little programming skill and deep learning knowledge can produce sophisticated fake videos using Deepfake. A number of indiscriminate fake videos has been increased significantly, which may lead to problems such as privacy violations, fake news and fraud. Therefore, it is necessary to detect fake video clips that cannot be discriminated by a human eyes. Thus, in this paper, we propose a deep-fake detection model applied with Bidirectional Convolution LSTM and Attention Module. Unlike LSTM, which considers only the forward sequential procedure, the model proposed in this paper uses the reverse order procedure. The Attention Module is used with a Convolutional neural network model to use the characteristics of each frame for extraction. Experiments have shown that the model proposed has 93.5% accuracy and AUC is up to 50% higher than the results of pre-existing studies.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.25 no.2
• /
• pp.234-239
• /
• 2021

Recently, as the installation of a large number of high-performance CCTVs for crime prevention and traffic control has increased rapidly, the problem of increasing system requirements for response to privacy infringement factors and analysis of high-definition image information transmitted from multiple cameras has been actively emerging. Accordingly, there is a need for a method for responding to privacy infringement and a method for efficiently processing surveillance images input from multiple cameras. In this paper, in order to reduce the processing cost of the input image and improve the processing speed, an integrated image is generated by grouping images input from a plurality of cameras. After analyzing the pre-generated integrated video, it detects a preset privacy event or an event that highlights interest. Depending on whether or not an event is detected, you will perform an editing operation corresponding to the event.

• Journal of the Korea Society of Computer and Information
• /
• v.27 no.1
• /
• pp.1-8
• /
• 2022

In this paper, we propose a fire detection system based on CCTV images using an object tracking technology with YOLOv4 model capable of real-time object detection and a DeepSORT algorithm. The fire detection model was learned from 10800 pieces of learning data and verified through 1,000 separate test sets. Subsequently, the fire detection rate in a single image and fire detection maintenance performance in the image were increased by tracking the detected fire area through the DeepSORT algorithm. It is verified that a fire detection rate for one frame in video data or single image could be detected in real time within 0.1 second. In this paper, our AI fire detection system is more stable and faster than the existing fire accident detection system.

• Journal of Convergence for Information Technology
• /
• v.12 no.1
• /
• pp.15-21
• /
• 2022

In the content-based video retrieval applications, the information on the movement of an object can be used as important in classifying the content. In particular, analyzing and classifying human movement can be used for various purposes as well as retrieval. In this paper, a method to improve the performance of the shape variation descriptor and shape sequence to describe and classify movement using shape information that changes according to the movement of an object is proposed. By selecting a shape descriptor to more efficiently describe the shape information of an object and comparing the distance function used to measure the similarity, the description and retrieval efficiency of movement information can be increased. Through experiments, it was shown that the proposed method can describe movement information more efficiently and increase the retrieval efficiency compared to the previous method.