이동물체 탐지 및 추적을 위한 에너지 보정 스네이크(ECS) 알고리즘의 실험 및 평가

윤희병,양성실 한국정보처리학회 2009 정보처리학회논문지. 소프트웨어 및 데이터 공학 Vol.16 No.4

Active Contour Model, that is, Snake algorithm is effective for detection and tracking the objects. However, this algorithm has some drawbacks; numerous parameters must be designed(weighting factors, iteration steps, etc.), a reasonable initialization must be available and moreover suffers from numerical instability. Therefore we propose a novel Energy Corrected Snake(ECS) algorithm which improved on external energy of Snake algorithm for detection and tracking the moving object more effectively. The proposed algorithm uses the difference image, getting when the object is moving. It copies four direction images from the difference image and performs the accumulating compute to erasing image noise, so that it gets external energy steadily. Then external energy united with contour that is computed by internal energy. Consequently we can detect and track the moving object more speedily and easily. To show the effectiveness of the proposed algorithm, we experiment on 3 situations. The experimental results showed that the proposed algorithm outperformed by 6~9% of detection rate and 6~11% of tracker detection rate compared with the Snake algorithm. 능동 윤곽선 모델, 즉 스네이크 알고리즘은 물체 탐지 및 추적에 사용되는 유용한 알고리즘이다. 그러나 이 알고리즘은 요소별 가중치 부여 및 반복단계 시 많은 변수가 필요하고, 초기화 애로 및 계산상 불안정성 등의 단점이 있다. 따라서 본 논문에서는 이러한 단점을 개선하여 보다 효과적인 이동물체 탐지 및 추적을 위해 기존 스네이크 알고리즘의 외부 에너지를 개선한 새로운 에너지 보정 스네이크(ECS) 알고리즘을 제안한다. 이를 위해 이동물체 이동 시 획득한 차영상 이미지를 4개의 방향성 이미지로 복사하고 각 이미지 픽셀에 대해 누적 연산 후 에너지 강화배열 내 저장 및 노이즈 제거를 통해 안정적인 이미지, 즉 외부 에너지를 획득한다. 또한 별도로 계산된 내부 에너지를 통해 얻어진 윤곽선(contour)을 외부 에너지에 병합함으로써 빠르고 쉬운 이동물체 탐지 및 추적이 가능하다. 제안한 알고리즘의 효용성을 확인하기 위해 3가지 상황을 대상으로 실험하였다. 실험 결과, 제안한 알고리즘이 기존 스네이크 알고리즘에 비해 탐지율은 평균 6~9%, 추적율은 6~11% 정도의 향상을 보였다.

고속도로 돌발상황검지알고리즘 성능 개선기법에 관한 연구

강수구,손봉수,도철웅,이시복 대한교통학회 2001 대한교통학회지 Vol.19 No.6

혼잡은 반복정체와 비반복(돌발상황)정체로 나눌 수 있다. 고속도로의 효율적인 교통관리를 위해서는 사전 예측이 불가능한 돌발상황으로 인한 비 반복정체를 신속·정확하게 검지하여 대응함이 매우 중요하다. 이러한 목적으로 국내외적으로 많은 노력을 기울여 왔으나, 기존에 개발된 많은 돌발상황검지알고리즘들은 검지율 증가시 수반되는 오보율증가에 대한 문제점을 극복하는데 어려움을 겪고 있는 실정이다. 이로 인해 고속도로 운영자에게 실질적으로 적용이 가능한 돌발상황검지알고리즘을 제공하지 못하고 있다. 이와 같은 현 여건을 감안하여 본 연구에서는 돌발상황 검지율과 오보율을 모두 향상시킬 수 있는 기존에 개발된 방법론을 응용하고, 운영자측면에서 돌발상황검지의 효용성을 증대할 수 있는 기법을 개발하는데 주안점을 두고 있다. 본 연구에서 제안한 알고리즘은 교통량, 속도, 점유율 등 세 교통변수간 관계식에 근거하여 일차적으로 돌발상황을 검지하고, 다음 단계에서 검지된 교통상황의 돌발상황 여부를 McMaster 알고리즘에서 적용한 반복 및 비반복정체 구분을 위한 교통상황판단 방법론을 응용하여 구축하였다. 본 연구에서 제안한 알고리즘의 구성체계와 연구에 적용된 방법론의 적합성 평가를 위해 미국 캘리포니아에 위치한 I-880고속도로에서 수집된 교통자료를 이용하였다. 평가결과, 돌발상황 검지율과 오보율 측면에서 본 연구에서 제시한 알고리즘의 적용가능성이 매우 긍정적으로 판단된다. 본 연구에서는 캘리포니아 알고리즘에 대한 평가결과도 참고적으로 제시하였다. Incident detection rate and false alarm rate are the key measures tot estimating the performance of automatic incident detection algorithms. It is, however inherently very difficult to improve the two measures simultaneously. The main purpose of this study is to present some methods for solving the problem. For this, an incident detection algorithm has been designed in this study. The algorithm is consisted of two functions, one for detecting incident and another for detecting congestion. A logic for distinguishing non-recurrent congestion from recurrent congestion was employed in the algorithm. The new algorithm basically requires speed, flow, and occupancy data for defining incident situation, but the algorithm is able to perform this task without one of the three parameters. The performance of the algorithm has been evaluated by using the field data collected from Interstate Highway 880 in Bay Area, California. The empirical analysis results are very promising and thus, the algorithm proposed may be very useful for the analysts. This paper presents some empirical test results for the performance of California incident detection algorithm, only for the reference purpose.

나이브 베이즈 분류기를 이용한 돌발상황 검지 알고리즘 개발

강성관,권봉경,권철우,박상민,윤일수 한국ITS학회 2018 한국ITS학회논문지 Vol.17 No.6

The purpose of this study is to develop an efficient incident detection algorithm by applying machine learning, which is being widely used in the transport sector. As a first step, network of the target site was constructed with micro-simulation model. Secondly, data has been collected under various incident scenarios produced with combination of variables that are expected to affect the incident situation. And, detection results from both McMaster algorithm, a well known incident detection algorithm, and the Naive Bayes algorithm, developed in this study, were compared. As a result of comparison, Naive Bayes algorithm showed less negative effect and better detect rate (DR) than the McMaster algorithm. However, as DR increases, so did false alarm rate (FAR). Also, while McMaster algorithm detected in four cycles, Naive Bayes algorithm determine the situation with just one cycle, which increases DR but also seems to have increased FAR. Consequently it has been identified that the Naive Bayes algorithm has a great potential in traffic incident detection. 본 연구에서는 최근 활발하게 활용되고 있는 머신러닝 기법을 교통분야에 적용하여 효율적인 돌발상황 검지 알고리즘을 개발하는 것을 목적으로 하였다. 미시교통시뮬레이션 모형을 통하여 대상지의 네트워크를 구축하였고 돌발상황에 영향을 줄 것으로 예상되는 변수의 여러 조합을 통해 시나리오를 설정하여 가상의 돌발상황 데이터를 수집하였다. 다음으로 대표적인 돌발상황 검지 알고리즘인 McMaster 알고리즘과 본 연구에서 개발한 나이브 베이즈 분류기 를구현하여 비교·평가하였다. 비교 결과, 나이브 베이즈 분류기가 McMaster 알고리즘에 비해 돌발상황 검지 간격에 따른 부정적인 영향이 적었고 더 우수한 검지율을 보였다. 하지만 검지율이 증가하는 만큼 오검지율 또한 증가하는 것을 확인할 수 있었다. McMaster 알고리즘은 4주기를 통해 검지가 가능하지만 나이브 베이즈 분류기는 1주기(30초)만으로 돌발상황을 판단할 수있다. 본 연구를 통해 개발한 나이브 베이즈 분류기가 효율적으로 돌발을 파악할 수 있다는것을 확인할 수 있었다.

Industrial Flame Edge Detection Algorithm Based on Gray Dominant Filter

Zhenhua Wei,Jianqiang Qiao 보안공학연구지원센터 2015 International Journal of Multimedia and Ubiquitous Vol.10 No.2

Industrial furnace flame detection is directly related to the safety and economic operation of boiler. The flame edge detection is one of the most key parts of the flame detection. At present, there are many edge detection algorithms for normal object but few for industrial flame edge detection which cannot provide fully support for furnace flame combustion stability, flame 3D reconstruction and flame temperature field reconstruction. One new industrial flame image edge detection algorithm based on industrial flame image characteristics is proposed in this paper. This algorithm can effectively remove noise level in flame image and well maintain flame image edge features, compute the gradients from multiple-directions for entirely detecting flame edge information. Moreover, this algorithm is verified by experiment in this paper to present high performance in resisting noise, and clearly see continuous flame edge curve, which reveals better effect in industrial flame edge detection than other edge detection algorithms.

휴대용 심전도 측정장치를 위한 실시간 QRS-complex 검출 알고리즘 개발

안휘,심형진,박재순,임종태,정연호 대한의용생체공학회 2022 의공학회지 Vol.43 No.4

In this paper, we present a QRS-complex detection algorithm to calculate an accurate heartbeat and clearly recognize irregular rhythm from ECG signals. The conventional Pan-Tompkins algorithm brings false QRS detection in the derivative when QRS and noise signals have similar instant variation. The proposed algorithm uses amplitude differences in 7 adjacent samples to detect QRS-complex which has the highest amplitude variation. The calculated amplitude is cubed to dominate QRS-complex and the moving average method is applied to diminish the noise signal’s amplitude. Finally, a decision rule with a threshold value is applied to detect accurate QRS-complex. The calculated signals with Pan-Tompkins and proposed algorithms were compared by signal-to-noise ratio to evaluate the noise reduction degree. QRS-complex detection performance was confirmed by sensitivity and the positive predictive value(PPV). Normal ECG, muscle noise ECG, PVC, and atrial fibrillation signals were achieved which were measured from an ECG simulator. The signal-to-noise ratio difference between Pan-Tompkins and the proposed algorithm were 8.1, 8.5, 9.6, and 4.7, respectively. All ratio of the proposed algorithm is higher than the Pan-Tompkins values. It indi- cates that the proposed algorithm is more robust to noise than the Pan-Tompkins algorithm. The Pan-Tompkins algo- rithm and the proposed algorithm showed similar sensitivity and PPV at most waveforms. However, with a noisy atrial fibrillation signal, the PPV for QRS-complex has different values, 42% for the Pan-Tompkins algorithm and 100% for the proposed algorithm. It means that the proposed algorithm has superiority for QRS-complex detection in a noisy environment.

Research on an Improved Intrusion Detection Algorithm

Yue Liu,Mei-shan Li 보안공학연구지원센터 2016 International Journal of Security and Its Applicat Vol.10 No.11

First of all, the principle of neural network is discussed, and the traditional BP network learning algorithm and BP neural network adaptive learning algorithm are researched. Combining the advantages of two algorithms, the distributed neural network self-learning algorithm is proposed, which is a kind of intrusion detection algorithm using the method of distributed learning to optimize the BP neural network algorithm. Using this algorithm to study and test the network intrusion data, it solves the problem that directly using BP learning caused by the training sample size too large and difficult to convergence. At the same time, the sample training time is shortened, and the BP neural network classification accuracy is improved. Secondly, based on the research of the improved algorithm, this paper gives the specific steps of the algorithm, and uses the improved algorithm to establish mathematical model which is used to analyzing and forecasting. Compared with the traditional BP network learning algorithm and BP neural network adaptive learning algorithm, verify the effectiveness and feasibility of the improved algorithm. Finally, the algorithm is applied to intrusion detection. Through appropriate test method, use the sample data of this paper adopted to verify the example. Through the results of the testing data, it verifies the performance of the distributed neural network self-learning algorithm, and comes to the conclusion.

영역정보기반의 유전자알고리즘을 이용한 텍스트 후보영역 검출

오준택(Jun-Taek Oh),김욱현(Wook-Hyun Kim) 대한전자공학회 2008 電子工學會論文誌-SP (Signal processing) Vol.45 No.6

본 논문은 화소 단위의 정보가 아닌 분할된 영역들의 정보를 기반으로 유전자 알고리즘을 이용한 텍스트 후보영역 검출방안을 제안한다. 먼저, 영상분할을 수행하기 위해 색상별 화소분류와 비동질적인 군집의 감소를 위한 영역 단위의 재분류 알고 리즘을 수행한다. 색상별 화소분류에 이용되는 EWFCM(Entropy-based Weighted Fuzzy C-Means) 알고리즘은 공간정보를 추가한 개선된 FCM 알고리즘으로써, 잡음에 강건한 특징을 가진다. EWFCM 알고리즘에 의해 분류된 화소들의 군집정보를 기반으로 수행되는 영역 단위의 재분류는 화소나 군집 단위의 재분류에 비해 효과적으로 영상에 존재하는 비동질적인 군집들을 감소시킬 수 있다. 그리고 텍스트 후보영역 검출은 분할된 영역들로부터 추출한 방향성 에지 성분에 대한 분산값 및 에너지, 크기, 개수 등의 정보를 기반으로 유전자알고리즘에 의해 수행된다. 이는 화소 단위의 정보를 이용한 방법보다 더 명확한 텍스트 영역정보를 획득할 수 있으며, 향후 자동문자인식에서 좀 더 손쉽게 이용될 수 있다. 실험 결과 제안한 분할방법은 기존 방법이나 화소나 군집 기반의 재분류보다 좋은 결과를 보였으며, 텍스트 후보영역 검출에서도 화소 단위의 정보를 이용한 기존 방법보다 더 좋은 결과를 보여 제안방법의 유효성을 확인하였다. This paper proposes a new text candidate region detection method that uses genetic algorithm based on information of the segmented regions. In image segmentation, a classification of the pixels at each color channel and a reclassification of the region-unit for reducing inhomogeneous clusters are performed. EWFCM(Entropy-based Weighted C-Means) algorithm to classify the pixels at each color channel is an improved FCM algorithm added with spatial information, and therefore it removes the meaningless regions like noise. A region-based reclassification based on a similarity between each segmented region of the most inhomogeneous cluster and the other clusters reduces the inhomogeneous clusters more efficiently than pixel- and cluster-based reclassifications. And detecting text candidate regions is performed by genetic algorithm based on energy and variance of the directional edge components, the number, and a size of the segmented regions. The region information-based detection method can singles out semantic text candidate regions more accurately than pixel-based detection method and the detection results will be more useful in recognizing the text regions hereafter. Experiments showed the results of the segmentation and the detection. And it confirmed that the proposed method was superior to the existing methods.

Community Discovering Algorithm based on Nodes with Maximum Degree and Label Propagation

Li Shengli,Chen Deyun,Yao Yuanzhe 보안공학연구지원센터 2016 International Journal of Grid and Distributed Comp Vol.9 No.12

Label propagation algorithm (LPA algorithm) is gotten widely attention by its simplicity, rapidity and greater effectiveness. Aimed at the problem of label updating process is sensitive to the order of nodes in LPA algorithm, an improved algorithm is proposed in this paper. The algorithm starts from the nodes with maximum degree in the network, according to its neighbor nodes’ community similarity index to decide whether its label setting, and then completes label setting in the first round; On this basis, it continues to carry out the iteration in label propagation algorithm, completes the entire network of community structure detection. The experimental results show that the improved algorithm is slightly better than the original algorithm at running time and number of iterations, has higher robustness, prevents the occurrence of the trivial solution.

돌발상황 검지를 위한 Wavelet 기법의 적용성 평가 : 서울특별시 도시고속도로를 중심으로

김동선(Kim Dong Sun),백주현(Baek Joo Hyun),송기한(Song Ki Han),이성모(Rhee Sung Mo) 대한토목학회 2006 대한토목학회논문집 D Vol.26 No.4D

우리나라 대부분의 도로에서는 교통사고를 비롯한 기대하지 않았던 비반복적인 사건 즉, 돌발상황의 발생이 증가되고 있다. 돌발상황의 증가는 교통류의 정상적인 흐름을 방해하며 용량감소를 초래하여 교통혼잡을 야기시킨다. 본 연구에서는 도시고속도로 돌발상황을 검지하기 위한 방법으로 현재 전기공학을 비롯한 많은 공학분야에서 사용하고 있는 Wavelet 기법을 적용 하였다. 이 기법을 적용하기 위한 실제자료는 서울시 도시고속도로(내부순환로)의 검지기를 통해 수집된 자료 중 교통류의 주요 변수인 점유율 자료를 이용하였다. Wavelet 기법의 적용성을 평가하기 위한 비교 알고리즘으로 기존 돌발상황 검지 알고리즘 중 점유율 기반의 California 알고리즘과 Low-Pass Filtering 알고리즘을 이용하여 모형의 유의성을 검토해 보았다. Wavelet 기법의 적용결과 오경보율(False Alarm Rate)은 California 알고리즘 및 Low-Pass Filtering 알고리즘과 비슷하였으나 검지율(Detection Rate) 측면에서는 다른 두 알고리즘보다 우수한 것으로 분석되었다. Incidents, which is unexpected unusual events such as traffic accidents, have increased on the most roads in Korea. The obstruction of a fluent traffic flow occurred by incidents causes the traffic congestion and decreases the capacity. The Wavelet technique was applied to detect the road section and the happening time of incidents on urban freeways in this study, and this technique has been widely used in many engineering fields such as an electrical engineering, etc. The availability and validity of the Wavelet technique to the detection of incidents was examined by the occupancy rate, the important element of traffic flows, which is extracted from the data of detectors installed on Seoul Urban freeways. Then, this result is compared to the California Algorithm and the Low-Pass Filtering Algorithm among basic present detection algorithms, which are based on the occupancy rate. Asa result, the false alarm rate of this method was similar as that of the California algorithm arid the Low-Pass Filtering algorithm" but the detection rate is higher.

A Fast and Accurate Algorithm for Detecting Community Structure in Social Networks

Junheng Huang,Yushan Sun,Yang Liu,Bailing Wang 보안공학연구지원센터(IJUNESST) 2015 International Journal of u- and e- Service, Scienc Vol.8 No.1

Studies of community structure and evolution in large social networks require fast and accurate algorithms for community detection. Among the existing algorithms for community detection, the label propagation algorithm (LPA) and the Newman modularity Q algorithm (NMA) have been widely used and studied in the community detection in large social networks, since the LPA has the advantages of near-linear running time, easy implementation and without requiring parameters, and the NMA is a relatively fast algorithm and has a clear metrics to measure community structure. However, the LPA has the shortcomings that the result of the community detection is instable and has a low quality. At the same time, disadvantages of the NMA are that it bases its decisions on purely local information about individual communities and gets the local optimal solution. In this paper, combined with these two algorithms, we propose a new community detection algorithm (LP-NMA), which extends the above two algorithms (the LPA and the NMA is a special case of the new algorithm respectively). The new algorithm not only retains the advantages of these two algorithms, but also has improved the stability and quality of community detection. Experiments on real social networks have proved that this method is better than the original LPA and NMA.