화학오염운 탐지를 위한 접촉식 화학탐지기를 탑재한 무인기와 원거리 화학탐지기의 복합 운용개념 고찰

이명재,정유진,정영수,이재환,남현우,박명규 한국군사과학기술학회 2020 한국군사과학기술학회지 Vol.23 No.3

Early-detection and monitoring of toxic chemical gas cloud with chemical detector is essential for reducing the number of casualties. Conventional method for chemical detection and reconnaissance has the limitation in approaching to chemically contaminated site and prompt understanding for the situation. Stand-off detector can detect and identify the chemical gas at a long distance but it cannot know exact distance and position. Chemical detection UAV is an emerging platform for its high mobility and operation safety. In this study, we have conducted chemical gas cloud detection with the stand-off chemical detector and the chemical detection UAV. DMMP vapor was generated in the area where the cloud can be detected through the field of view(FOV) of stand-off chemical detector. Monitoring the vapor cloud with standoff detector, the chemical detection UAV moved back and forth at the area DMMP vapor being generated to detect the chemical contamination. The hybrid detection system with standoff cloud detection and point detection by chemical sensors with UAV seems to be very efficient as a new concept of chemical detection.

경량화된 임베디드 시스템에서 역 원근 변환 및 머신 러닝 기반 차선 검출

홍성훈,박대진 대한임베디드공학회 2022 대한임베디드공학회논문지 Vol.17 No.1

This paper proposes a novel lane detection algorithm based on inverse perspective transformation and machine learning in lightweight embedded system. The inverse perspective transformation method is presented for obtaining a bird’s-eye view of the scene from a perspective image to remove perspective effects. This method requires only the internal and external parameters of the camera without a homography matrix with 8 degrees of freedom (DoF) that maps the points in one image to the corresponding points in the other image. To improve the accuracy and speed of lane detection in complex road environments, machine learning algorithm that has passed the first classifier is used. Before using machine learning, we apply a meaningful first classifier to the lane detection to improve the detection speed. The first classifier is applied in the bird’s-eye view image to determine lane regions. A lane region passed the first classifier is detected more accurately through machine learning. The system has been tested through the driving video of the vehicle in embedded system. The experimental results show that the proposed method works well in various road environments and meet the real-time requirements. As a result, its lane detection speed is about 3.85 times faster than edge-based lane detection, and its detection accuracy is better than edge-based lane detection. This paper proposes a novel lane detection algorithm based on inverse perspective transformation and machine learning in lightweight embedded system. The inverse perspective transformation method is presented for obtaining a bird’s-eye view of the scene from a perspective image to remove perspective effects. This method requires only the internal and external parameters of the camera without a homography matrix with 8 degrees of freedom (DoF) that maps the points in one image to the corresponding points in the other image. To improve the accuracy and speed of lane detection in complex road environments, machine learning algorithm that has passed the first classifier is used. Before using machine learning, we apply a meaningful first classifier to the lane detection to improve the detection speed. The first classifier is applied in the bird’s-eye view image to determine lane regions. A lane region passed the first classifier is detected more accurately through machine learning. The system has been tested through the driving video of the vehicle in embedded system. The experimental results show that the proposed method works well in various road environments and meet the real-time requirements. As a result, its lane detection speed is about 3.85 times faster than edge-based lane detection, and its detection accuracy is better than edge-based lane detection.

Personness estimation for real-time human detection on mobile devices

Kim, Kyuwon,Oh, Changjae,Sohn, Kwanghoon Elsevier 2017 expert systems with applications Vol.72 No.-

<P><B>Abstract</B></P> <P>One aim of detection proposal methods is to reduce the computational overhead of object detection. However, most of the existing methods have significant computational overhead for real-time detection on mobile devices. A fast and accurate proposal method of human detection called personness estimation is proposed, which facilitates real-time human detection on mobile devices and can be effectively integrated into part-based detection, achieving high detection performance at a low computational cost. Our work is based on two observations: (i) normed gradients, which are designed for generic objectness estimation, effectively generate high-quality detection proposals for the person category; (ii) fusing the normed gradients with color attributes improves the performance of proposal generation for human detection. Thus, the candidate windows generated by the personness estimation will very likely contain human subjects. The human detection is then guided by the candidate windows, offering high detection performance even when the detection task terminates prior to completion. This interruptible detection scheme, called anytime detection, enables real-time human detection on mobile devices. Furthermore, we introduce a new evaluation methodology called time-recall curves to practically evaluate our approach. The applicability of our proposed method is demonstrated in extensive experiments on a publicly available dataset and a real mobile device, facilitating acquisition and enhancement of portrait photographs (e.g. selfie) on widespread mobile platforms.</P> <P><B>Highlights</B></P> <P> <UL> <LI> A fast and accurate detection proposal method for the person category is proposed. </LI> <LI> Detection proposals are used by the part-based human detector in a improved way. </LI> <LI> High effectiveness of the proposed method is demonstrated on a real mobile device. </LI> </UL> </P>

Fast and Efficient Method for Fire Detection Using Image Processing

Turgay Celik 한국전자통신연구원 2010 ETRI Journal Vol.32 No.6

Conventional fire detection systems use physical sensors to detect fire. Chemical properties of particles in the air are acquired by sensors and are used by conventional fire detection systems to raise an alarm. However, this can also cause false alarms; for example, a person smoking in a room may trigger a typical fire alarm system. In order to manage false alarms of conventional fire detection systems, a computer vision-based fire detection algorithm is proposed in this paper. The proposed fire detection algorithm consists of two main parts: fire color modeling and motion detection. The algorithm can be used in parallel with conventional fire detection systems to reduce false alarms. It can also be deployed as a stand-alone system to detect fire by using video frames acquired through a video acquisition device. A novel fire color model is developed in CIE L*a*b* color space to identify fire pixels. The proposed fire color model is tested with ten diverse video sequences including different types of fire. The experimental results are quite encouraging in terms of correctly classifying fire pixels according to color information only. The overall fire detection system’s performance is tested over a benchmark fire video database, and its performance is compared with the state-of-the-art fire detection method.

데이터 마이닝을 이용한 공격 탐지 메커니즘의 실험적 비교 연구

김미희,오하영,채기준,Kim, Mi-Hui,Oh, Ha-Young,Chae, Ki-Joon 한국통신학회 2006 韓國通信學會論文誌 Vol.31 No.2c

본 논문에서는 최신의 공격 유형을 잘 분류해 내고, 기존 공격의 변형이나 새로운 공격에도 탐지 가능하도록 데이터 마이닝 기법을 이용한 공격 탐지 모델 생성 방법들을 소개하고, 다양한 실험을 통해 탐지율 및 탐지 시간 측면에서 이 모델들의 성능을 비교한다. 이러한 탐지 모델을 생성하는데 중요한 요소로 데이터, 속성, 탐지 알고리즘을 꼽을 수 있는데, 실제 네트워크에서 수집된 NetFlow 데이터와 대량의 KDD Cup 1999 데이터를 사용하였다. 또한 탐지 알고리즘으로서 단일 지도/비지도학습 데이터 마이닝 기법 및 결합된 방법을 이용하여 탐지 모델을 생성, 비교 실험하였다. 시험 결과, 결합된 지도학습 알고리즘을 사용한 경우 모델링 시간은 길었지만 가장 탐지율이 높았고, 모든 경우 탐지 시간이 1초 내외로 실시간 탐지 가능성을 입증할 수 있었다. 또한 새로운 공격에 대한 이상탐지 결과로도 92$\%$ 이상의 탐지율을 보임으로 탐지 가능성을 입증할 수 있었고, SOM 기법을 사용하는 경우에는 새로운 공격이 기존 어느 공격에 유사한 특성을 갖는지에 대한 부과적인 정보도 제공하였다. In this paper, we introduce the creation methods of attack detection model using data mining technologies that can classify the latest attack types, and can detect the modification of existing attacks as well as the novel attacks. Also, we evaluate comparatively these attack detection models in the view of detection accuracy and detection time. As the important factors for creating detection models, there are data, attribute, and detection algorithm. Thus, we used NetFlow data gathered at the real network, and KDD Cup 1999 data for the experiment in large quantities. And for attribute selection, we used a heuristic method and a theoretical method using decision tree algorithm. We evaluate comparatively detection models using a single supervised/unsupervised data mining approach and a combined supervised data mining approach. As a result, although a combined supervised data mining approach required more modeling time, it had better detection rate. All models using data mining techniques could detect the attacks within 1 second, thus these approaches could prove the real-time detection. Also, our experimental results for anomaly detection showed that our approaches provided the detection possibility for novel attack, and especially SOM model provided the additional information about existing attack that is similar to novel attack.

다양한 화소기반 변화탐지 결과와 등록오차를 이용한 객체기반 변화탐지

정세정,김태헌,이원희,한유경 한국측량학회 2019 한국측량학회지 Vol.37 No.6

Change detection, one of the main applications of multi-temporal satellite images, is an indicator that directly reflects changes in human activity. Change detection can be divided into pixel-based change detection and object-based change detection. Although pixel-based change detection is traditional method which is mostly used because of its simple algorithms and relatively easy quantitative analysis, applying this method in VHR (Very High Resolution) images cause misdetection or noise. Because of this, pixel-based change detection is less utilized in VHR images. In addition, the sensor of acquisition or geographical characteristics bring registration noise even if co-registration is conducted. Registration noise is a barrier that reduces accuracy when extracting spatial information for utilizing VHR images. In this study object-based change detection of VHR images was performed considering registration noise. In this case, object-based change detection results were derived considering various pixel-based change detection methods, and the major voting technique was applied in the process with segmentation image. The final object-based change detection result applied by the proposed method was compared its performance with other results through reference data. 다시기 위성 영상을 이용한 변화탐지 분석은 인간 활동의 변화를 직접 반영하는 지표이다. 변화탐지는 크게 화소 기반 변화탐지(PBCD: Pixel-Based Change Detection)와 객체 기반 변화탐지(OBCD: Object-Based Change Detection)로 구분한다. 화소 기반 변화탐지는 알고리즘이 간단하고 비교적 쉽게 정량적 분석이 가능해 전통적으로 많이 쓰여온 기법이나 고해상도 영상에서의 화소 기반 변화탐지는 오탐지나 노이즈(noise)가 발생하기 때문에 고해상도 영상에서의 활용도가 떨어진다. 또한, 고해상도 다시기 영상은 취득 당시 센서의 자세나 지형적 특성으로 인해 영상 등록(image registration)을 수행한 이후에도 지형적 불일치가 발생한다. 등록오차(registration noise)라고 불리는 이 지형 불일치는 고해상도 다시기 영상 활용을 위한 공간정보 추출 시 정확도를 떨어뜨리는 방해요인으로 작용한다. 이에 본 연구에서는 등록오차를 고려한 고해상도 영상의 객체 기반 변화탐지를 수행하였다. 이 때, 다양한 화소 기반 변화탐지 결과를 모두 고려한 객체 기반 변화탐지 결과를 도출하였으며 이 과정에서 분할 영상(segmentation image)과의 major voting을 적용하였다. 제안 기법과 화소 기반 변화탐지 결과, 그리고 화소 기반 변화탐지 결과를 객체 기반 변화탐지로 확장한 결과의 비교를 통해 제안 기법의 우수성을 평가하였다

Systematic Change Detection With Spectral Similarity Measures of SID for Uranium Tailing Piles to Monitor Suspicious Mining Activities in the Pyongsan Uranium Mine

Hoseong Choi,Gayeon Ha,Minsoo Kim 한국방사성폐기물학회 2022 한국방사성폐기물학회 학술논문요약집 Vol.20 No.1

With the enhancement of the spatial resolution of satellite imagery (1 m or less), the satellite image analysis has been considered as the indispensable means for remote sensing of nuclear proliferation activities in the restricted access areas such as North Korea. Notably, in the case of an open-pit uranium mine, e.g. the Pyongsan uranium mine, the mining activity can be presumed if detecting the location and extent uranium tailing piles near shafts within temporal images. Several studies have researched on the target detection for minerals of interest such as limestone and coal to evaluate the economic activities by utilizing similarity measures, e.g., a spectral angle mapper and a spectral information divergence (SID). Thus, this paper presented a systematic change detection methodology for monitoring the uranium mining activity in the Pyongsan uranium mine with a similarity measure of SID. The proposed methodology using the target detection results consists of the following five steps. The first step is to acquire stereo images of areas of interest for change detection. The second step is to preprocess the stereo images as following measures: (i) the QUick Atmospheric Correction and the image-to-image registration with ENVI and (ii) the Gram-Schmidt pansharpening. The third step is to extract spectral information for minerals of interest, i.e., uranium tailing piles, by sampling pixels within the reference image. It is based on the satellite analysis report for the Pyongsan uranium mine by CSIS, which specified the location of the uranium tailing piles. As the fourth step, the target detection for uranium tailing piles was performed through the similarity measure of SID between the extracted spectral information and the spectral reflectance of the image. In the fifth step, the change detection was processed using the multivariate alteration detection algorithm, which compares the target detection results by canonical correlation analysis. Furthermore, this paper evaluated the performance of the proposed methodology with the change detection accuracy assessment index, i.e., the area under a receiver operating characteristic curve. In conclusion, this paper suggests the systematic change detection methodology utilizing time series analysis of target detection for uranium tailing piles, which can save time and cost for humans to interpret large amounts of satellite information at the restricted access areas. As future works, the feasibility of the proposed methodology would be investigated by analyzing distribution of minerals of interest regarding nuclear proliferation at Yongbyon, which has the historical events of suspicious nuclear activities.

Zero Crossing Detection 회로 Modeling 및 응용연구

정성인 한국인터넷방송통신학회 2020 한국인터넷방송통신학회 논문지 Vol.20 No.4

교류 전압의 위상을 검출하여 제어하는 시스템의 경우 아날로그 제어방식에서는 검출한 위상에 대해 필터링에 의한 위상 offset 부분을 보상하여 제어에 응용하고 있다. 그러나, 디지털 제어방식에서는 이러한 위상 검출을 이용하여 제어할 경우 마이크로프로세서 혹은 마이크로 컨트롤러의 동작 주파수와 입력 위상 시간과의 오차로 인하여 정밀한 제어 를 이룰 수가 없다. 일반적으로 사용하는 방식이 일정한 시간이 되면 누적된 오차를 임의로 보상하여 맞추어주는 방식인 데 이러한 경우 보상하기 전까지는 오차를 지속적으로 가지고 갈 수밖에 없는 상황이 발생하게 된다. 이러한 문제점을 해결하기 위해서는 실시간으로 영점을 검출하여 마이크로프로세서의 동작 주파수에 맞도록 보상하는 방법이 필요하게 된다. 따라서 이러한 오차를 줄이면서 정밀한 디지털 제어에 응용하기 위해 본 논문에서 수행하고자 하는 연구는 다음과 같다. 1) 시뮬레이션 모델링을 통해 Zero Crossing Detection 알고리즘을 구현하여 영점을 검출을 통하여 동작 주파수 에 맞도록 보상하는 방법에 관해 연구. 2) Microcontroller를 이용한 Zero Crossing Detection 설계를 통하여 실시간 으로 영점을 검출하여 마이크로프로세서의 동작 주파수에 맞도록 보상하는 방법에 관해 연구. 3) Zero Crossing Detection 회로를 활용하여 BLDC 전동기의 회전자 위치 추정 연구. In the case of a system that detects and controls the phase of an alternating voltage, the analog control method compensates the phase offset part by filtering for the detected phase and applies it to the control. However, in the digital control method, precise control cannot be achieved due to an error between the operating frequency of the microprocessor or the microcontroller and the input phase time when controlled using such phase detection. In general, when the method used is a certain time, the accumulated error is compensated and adjusted at random. To solve this problem, a method of detecting a zero point in real time and compensating for the operating frequency of the microprocessor is needed. Therefore, the research to be performed in this paper to reduce these errors and apply them to precise digital control is as follows. 1) Research on how to implement Zero Crossing Detection algorithm through simulation modeling to compensate the zero point to match the operating frequency through detection. 2) A study on the method of detecting zero points in real time through the Zero Crossing Detection design using a microcontroller and compensating for the operating frequency of the microprocessor. 3) A study on the estimation of the rotor position of BLDC motors using the Zero Crossing Detection circuit.

A Mask Wearing Detection System Based on Deep Learning

Yang, Shilong,Xu, Huanhuan,Yang, Zi-Yuan,Wang, Changkun Korea Multimedia Society 2021 The journal of multimedia information system Vol.8 No.3

COVID-19 has dramatically changed people's daily life. Wearing masks is considered as a simple but effective way to defend the spread of the epidemic. Hence, a real-time and accurate mask wearing detection system is important. In this paper, a deep learning-based mask wearing detection system is developed to help people defend against the terrible epidemic. The system consists of three important functions, which are image detection, video detection and real-time detection. To keep a high detection rate, a deep learning-based method is adopted to detect masks. Unfortunately, according to the suddenness of the epidemic, the mask wearing dataset is scarce, so a mask wearing dataset is collected in this paper. Besides, to reduce the computational cost and runtime, a simple online and real-time tracking method is adopted to achieve video detection and monitoring. Furthermore, a function is implemented to call the camera to real-time achieve mask wearing detection. The sufficient results have shown that the developed system can perform well in the mask wearing detection task. The precision, recall, mAP and F1 can achieve 86.6%, 96.7%, 96.2% and 91.4%, respectively.

알려지지 않은 위협 탐지를 위한 CBA와 OCSVM 기반 하이브리드 침입 탐지 시스템

신건윤 ( Gun-yoon Shin ),김동욱 ( Dong-wook Kim ),윤지영 ( Jiyoung Yun ),김상수 ( Sang-soo Kim ),한명묵 ( Myung-mook Han ) 한국인터넷정보학회 2021 인터넷정보학회논문지 Vol.22 No.3

인터넷이 발달함에 따라, IoT, 클라우드 등과 같은 다양한 IT 기술들이 개발되었고, 이러한 기술들을 사용하여 국가와 여러 기업들에서는 다양한 시스템을 구축하였다. 해당 시스템들은 방대한 양의 데이터들을 생성하고, 공유하기 때문에 시스템에 들어있는 중요한 데이터들을 보호하기 위해 위협을 탐지할 수 있는 다양한 시스템이 필요하였으며, 이에 대한 연구가 현재까지 활발히 진행되고 있다. 대표적인 기술로 이상 탐지와 오용 탐지를 들 수 있으며, 해당 기술들은 기존에 알려진 위협이나 정상과는 다른 행동을 보이는 위협들을 탐지한다. 하지만 IT 기술이 발전함에 따라 시스템을 위협하는 기술들도 점차 발전되고 있으며, 이러한 탐지 방법들을 피해서 위협을 가한다. 지능형 지속 위협(Advanced Persistent Threat : APT)은 국가 또는 기업의 시스템을 공격하여 중요 정보 탈취 및 시스템 다운 등의 공격을 수행하며, 이러한 공격에는 기존에 알려지지 않았던 악성코드 및 공격 기술들을 적용한 위협이 존재한다. 따라서 본 논문에서는 알려지지 않은 위협을 탐지하기 위한 이상 탐지와 오용 탐지를 결합한 하이브리드 침입 탐지 시스템을 제안한다. 두 가지 탐지 기술을 적용하여 알려진 위협과 알려지지 않은 위협에 대한 탐지가 가능하게 하였으며, 기계학습을 적용함으로써 보다 정확한 위협 탐지가 가능하게 된다. 오용 탐지에서는 Classification based on Association Rule(CBA)를 적용하여 알려진 위협에 대한 규칙을 생성하였으며, 이상 탐지에서는 One Class SVM(OCSVM)을 사용하여 알려지지 않은 위협을 탐지하였다. 실험 결과, 알려지지 않은 위협 탐지 정확도는 약 94%로 나타난 것을 확인하였고, 하이브리드 침입 탐지를 통해 알려지지 않은 위협을 탐지 할 수 있는 것을 확인하였다. With the development of the Internet, various IT technologies such as IoT, Cloud, etc. have been developed, and various systems have been built in countries and companies. Because these systems generate and share vast amounts of data, they needed a variety of systems that could detect threats to protect the critical data contained in the system, which has been actively studied to date. Typical techniques include anomaly detection and misuse detection, and these techniques detect threats that are known or exhibit behavior different from normal. However, as IT technology advances, so do technologies that threaten systems, and these methods of detection. Advanced Persistent Threat (APT) attacks national or companies systems to steal important information and perform attacks such as system down. These threats apply previously unknown malware and attack technologies. Therefore, in this paper, we propose a hybrid intrusion detection system that combines anomaly detection and misuse detection to detect unknown threats. Two detection techniques have been applied to enable the detection of known and unknown threats, and by applying machine learning, more accurate threat detection is possible. In misuse detection, we applied Classification based on Association Rule(CBA) to generate rules for known threats, and in anomaly detection, we used One-Class SVM(OCSVM) to detect unknown threats. Experiments show that unknown threat detection accuracy is about 94%, and we confirm that unknown threats can be detected.