• Proceedings of the Korean Society of Computer Information Conference
• /
• 2022.07a
• /
• pp.81-84
• /
• 2022

제조 산업에서의 이상치 검출은 생산품의 품질과 운영비용을 절감하기 위한 중요한 요소로 최근 딥러닝을 사용하여 자동화되고 있다. 이상치 검출을 위한 딥러닝 기법에는 CNN이 있으며, CNN을 계층적으로 구성할 경우 단일 CNN 모델에 비해 상대적으로 성능의 향상을 보일 수 있다는 것이 많은 선행 연구에서 나타났다. 이에 MVTec-AD 데이터셋을 이용하여 계층 CNN이 다중 클래스 이상치 판별 문제에 대해 효과적인지를 탐구하고자 하였다. 실험 결과 단일 CNN의 정확도는 0.7715, 계층 CNN의 정확도는 0.7838로 다중 클래스 이상치 판별 문제에 있어 계층 CNN 방식 접근이 다중 클래스 이상치 탐지 문제에서 알고리즘의 성능을 향상할 수 있음을 확인할 수 있었다. 계층 CNN은 모델과 파라미터의 개수와 리소스의 사용이 단일 CNN에 비하여 기하급수적으로 증가한다는 단점이 존재한다. 이에 계층 CNN의 장점을 유지하며 사용 리소스를 절약하고자 하였고 K-means, GMM, 계층적 클러스터링 알고리즘을 통해 제작한 새로운 클래스를 이용해 계층 CNN을 구성하여 각각 정확도 0.7930, 0.7891, 0.7936의 결과를 얻을 수 있었다. 이를 통해 Clustering 알고리즘을 사용하여 적절히 물체를 분류할 경우 물체에 따른 개별 상태 판단 모델을 제작하는 것과 비슷하거나 더 좋은 성능을 내며 리소스 사용을 줄일 수 있음을 확인할 수 있었다.

• Korean Journal of Radiology
• /
• v.24 no.11
• /
• pp.1061-1080
• /
• 2023

Artificial intelligence (AI) in radiology is a rapidly developing field with several prospective clinical studies demonstrating its benefits in clinical practice. In 2022, the Korean Society of Radiology held a forum to discuss the challenges and drawbacks in AI development and implementation. Various barriers hinder the successful application and widespread adoption of AI in radiology, such as limited annotated data, data privacy and security, data heterogeneity, imbalanced data, model interpretability, overfitting, and integration with clinical workflows. In this review, some of the various possible solutions to these challenges are presented and discussed; these include training with longitudinal and multimodal datasets, dense training with multitask learning and multimodal learning, self-supervised contrastive learning, various image modifications and syntheses using generative models, explainable AI, causal learning, federated learning with large data models, and digital twins.

• Journal of the Korea Society for Simulation
• /
• v.18 no.3
• /
• pp.73-82
• /
• 2009

Due to the many advantages including low price, low power consumption, and miniaturization, the CMOS camera has been utilized in many applications, including mobile phones, the automotive industry, medical sciences and sensoring, robotic controls, and research in the security field. In particular, the 360 degree omni-directional camera when utilized in multi-camera applications has displayed issues of software nature, interface communication management, delays, and a complicated image display control. Other issues include energy management problems, and miniaturization of a multi-camera in the hardware field. Traditional CMOS camera systems are comprised of an embedded system that consists of a high-performance MCU enabling a camera to send and receive images and a multi-layer system similar to an individual control system that consists of the camera's high performance Micro Controller Unit. We proposed the SL-AVS (Small Size/Low power Around-View System) to be able to control a camera while collecting image data using a high speed synchronization technique on the foundation of a single layer low performance MCU. It is an initial model of the omni-directional camera that takes images from a 360 view drawing from several CMOS camera utilizing a 110 degree view. We then connected a single MCU with four low-power CMOS cameras and implemented controls that include synchronization, controlling, and transmit/receive functions of individual camera compared with the traditional system. The synchronization of the respective cameras were controlled and then memorized by handling each interrupt through the MCU. We were able to improve the efficiency of data transmission that minimizes re-synchronization amongst a target, the CMOS camera, and the MCU. Further, depending on the choice of users, respective or groups of images divided into 4 domains were then provided with a target. We finally analyzed and compared the performance of the developed camera system including the synchronization and time of data transfer and image data loss, etc.

• Korean Security Journal
• /
• no.58
• /
• pp.9-34
• /
• 2019

The Korean Police Force is equipped with the dual responsibility to not only protect the constitutional right to protest, but also prevent potential disorder and misconduct might be caused by the abuse of such a right. To this end, the Korean national police employ the crowd counting methodology, termed 'Maximum Figure at Any One Time' with a view to dispatching the proportionate number of police officers to protest scenes for safety management. However, protest organizers rather take advantage of 'Cumulative Figure' methodology, the purpose of which being to publicize the wide recognition of success, noticeably by demonstrating that as many people as possible support for their cause or voice. Hence, different estimates generated by different methods have raised serious political issues in Korean society. Nevertheless, it is found out that there are only three existing academic studies in Korea regarding crowd counting methods, and they are mainly geared towards comparing the two methods, unfortunately without any attempt to analyze the foreign literature in details. Keeping the research gap in mind, the research conducts a comprehensive review of the foreign literature with relation to protest crowd counting methods. Derived from the review and analysis, the counting methods can be broadly categorized into the three models such as: 1) Grid/Density Model, 2) Moving Crowds Model, and 3) Electronic & Non-Image Model. In the end, the research provides brief explanations regarding specific research findings per each model, and further, suggests some policy implications for the development of more accurate crowd counting methodology at protests in Korea.

• Journal of Internet Computing and Services
• /
• v.21 no.4
• /
• pp.17-23
• /
• 2020

Biometric information indicating measurement items related to human characteristics has attracted great attention as security technology with high reliability since there is no fear of theft or loss. Among these biometric information, fingerprints are mainly used in fields such as identity verification and identification. If there is a problem such as a wound, wrinkle, or moisture that is difficult to authenticate to the fingerprint image when identifying the identity, the fingerprint expert can identify the problem with the fingerprint directly through the preprocessing step, and apply the image processing algorithm appropriate to the problem. Solve the problem. In this case, by implementing artificial intelligence software that distinguishes fingerprint images with cuts and wrinkles on the fingerprint, it is easy to check whether there are cuts or wrinkles, and by selecting an appropriate algorithm, the fingerprint image can be easily improved. In this study, we developed a total of 17,080 fingerprint databases by acquiring all finger prints of 1,010 students from the Royal University of Cambodia, 600 Sokoto open data sets, and 98 Korean students. In order to determine if there are any injuries or wrinkles in the built database, criteria were established, and the data were validated by experts. The training and test datasets consisted of Cambodian data and Sokoto data, and the ratio was set to 8: 2. The data of 98 Korean students were set up as a validation data set. Using the constructed data set, five CNN-based architectures such as Classic CNN, AlexNet, VGG-16, Resnet50, and Yolo v3 were implemented. A study was conducted to find the model that performed best on the readings. Among the five architectures, ResNet50 showed the best performance with 81.51%.

• Journal of the Korean Society of Radiology
• /
• v.6 no.2
• /
• pp.143-149
• /
• 2012

Today each hospital is trend that change rapidly by up to date, digitization and introducing newest medical treatment equipment. So, we introduce new CR system and supplement film system's shortcoming and PACS, EMR, RTP system's network that is using in hospital harmoniously and accomplish quality improvement of medical treatment and service elevation about business efficiency enlargement and patient Accordingly, we wish to introduce our case that integrate reflex that happen with radiation oncology here upon to PACS using CR system and estimate the availability. We measured that is Gantry, Collimator Star Shot, Light vs. Radiation, HDR QA(Dwell position accuracy) with Medical LINAC(MEVATRON-MX) Then, PACS was implemented on the digital images on the monitor that can be confirmed through the QA. Also, for cooperation with OCS system that is using from present source and impose code that need in treatment in each treatment, did so that Order that connect to network, input to CR may appear, did so that can solve support data mistake (active Pinacle's case supports DICOM3 file from present source but PACS does not support DICOM3 files.) of Pinacle and PACS that is Planning System and look at Planning premier in PACS. All image and data constructed integration to PACS as can refer and conduct premier in Hospital anywhere using CR system. Use Dosimetry IP in Filmless environment and QA's trial such as Light/Radition field size correspondence, gantry rotation axis' accuracy, collimator rotation axis' accuracy, brachy therapy's Dwell position check is available. Business efficiency by decrease and so on of unnecessary human strength consumption was augmented accordingly with session shortening as that integrate premier that is neted with radiation oncology using CR system to PACS. and for the future patient information security is essential.

• Journal of Legislation Research
• /
• no.44
• /
• pp.725-756
• /
• 2013

In this paper, the possibility of extension of falsification conception was investigated to discuss the validity of this precedent. Consequently this extension was indispensible according to the article 237-2 of criminal code which regards the reproduced document as the document in the document crime. However, this is against the security of human right. On the contrary, there is not this kind of article in the German criminal code and the German precedents and majority theory are negative to regard it as the criminal document. And also, there is the pont of view that the reproduced document is not the criminal document because it's not the expression itself of document nominee's intention, so the article 237-2 should be demolished in Korea. According to this opinion, the serious reconsideration should be required in the legislation of this article 237-2. Nevertheless, if this extended conception is needed and it's possible, the meaning of the computer monitor's image is not able to be ignored in the conception of falsification. Therefore this should be regarded as the element of the falsification conception. In other words, this can't be evaluated as the object of falsification but the target of falsification, according to the conception extension, though the precedents do not regard it as the document in document crime.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.15 no.6
• /
• pp.714-719
• /
• 2005

The paper attempts to design and implement the system which can remotely check visitors' speech or Image by a mobile phone. This system is designed to recognize who a visitor is through the automatic calling service, not through a short message, via the mobile phone, even when the home owner is outside. In general, door locks are controlled through the home Server, but it is more effective to control door locks by using DTMF signal from a real-time point of view. The technology suggested in this paper makes it possible to communicate between the visiter and the home owner by making a phone call to tile home owner's mobile phone automatically when the visiter visits the house even if the home owner is outside, and if necessary, it allows for the home owner to control the door lock remotely. Thanks to the system, the home owner is not restricted by time or space for checking the visitor's identification and controlling the door lock. In addition, the security system is improved by changing from the existing password form to the combination of password and speaker verification lot the verification procedure required for controlling the door lock and setting the environment under consideration of any disadvantages which may occur when the mobile Phone is lost. Also, any existing problems such as reconnection to tile network for controlling tile door lock are solved by controlling the door lock in real time by use of DTMF signal while on the phone.

• Journal of the Korean Institute of Telematics and Electronics S
• /
• v.36S no.2
• /
• pp.69-80
• /
• 1999

Recently, social requirements of personal identification techniques are rapidly expanding in a number of new application ares. Especially fingerprint recognition is the most important technology. Fingerprint recognition technologies are well established, proven, cost and legally accepted. Therefore, it has more spot lighted among the any other biometrics technologies. In this paper we propose a new on-line fingerprint recognition algorithm for non-inked type live scanner to fit their increasing of security level under the computing environment. Fingerprint recognition system consists of two distinct structural blocks: feature extraction and feature matching. The main topic in this paper focuses on the feature matching using the fingerprint minutiae (ridge ending and bifurcation). Minutiae matching is composed in the alignment stage and matching stage. Success of optimizing the alignment stage is the key of real-time (on-line) fingerprint recognition. Proposed alignment algorithm using clique shows the strength in the search space optimization and partially incomplete image. We make our own database to get the generality. Using the traditional statistical discriminant analysis, 0.05% false acceptance rate (FAR) at 8.83% false rejection rate (FRR) in 1.55 second average matching speed on a Pentium system have been achieved. This makes it possible to construct high performance fingerprint recognition system.

• Journal of Korean Society for Geospatial Information Science
• /
• v.21 no.2
• /
• pp.3-9
• /
• 2013

Various data like digital maps(1/1,000 or 1/5,000), field surveying, online materials and literatures are used for the preliminary investigation for urban development such as the feasibility evaluation, the profitability analysis, the zoning proposal, the zoning designation, and the land replotting planning. There are a couple of urban development methods like an expropriation, a replotting, a mixed-used method. The replotting method requires the consideration of land replotting types based on topography and building condition, which is not easy to gather data for the preliminary investigation maintaining the security of development planning. There are limitations of a preliminary investigation using aerial photos to detect topographic and building changes at specific period. GIS data combined with high-resolution imagery has advantages over the current dataset, which come from easy acquisition of various spatial resolution satellite images, wide swath coverage, the choice of imagery resolution satisfying a usage purpose, economic benefit comparing to aerial photos, and the calculation of distance and area on imagery from image modeling. For these reasons, the proposed method in this study enables to perform the more appropriate preliminary investigation using more accurate information.