基於攝影機管理的異常檢測機制

Anomaly Detection Mechanism Based on Camera Management

簡哲毅(Jhe-Yi Jian) ； 黃揚清(Yang-Ching Huang) ； 林于淵(Yu-Yuan Lin)

《先進工程學刊》 17卷3期 (2022/07) Pp. 41-47

摘要

廣域監控的場景中例如城市交通、鐵路、軍營、科技園區、等重要場域，往往都有部署大量的攝影機，其中在道路監控場域是規模最大數量最多的，且需全時運作，因此如何有效維護設備變成一個非常重要的議題。以往都是靠人工方式定期查修攝影機，這是非常耗時以及人力，在發現攝影機故障的時候，都是已丟失關鍵的影像，為了提升異常檢測效率並降低人力，本文提出三種攝影機檢測異常方法。透過影像管理主機定期擷取攝影機快照以及收集錄影檔時間紀錄可以分析出攝影機連線與錄影狀態，當有發現斷線或是異常情形系統就可以進行報修處理，此外透過攝影機快照可以使用AI模糊辨識，標記出畫面模糊或是失焦的攝影機。

關鍵字

監控攝影機 ； 異常檢測 ； 深度學習 ； 模糊偵測

並列摘要

In the scene of wide area monitoring, there are often a large number of surveillance cameras, such as urban transportation, railway, military camp, science park and other important areas. Among them, the intersection field monitoring requires full-time operation and it has a largest number on a massive scale. Therefore, how to effectively maintain the device has become a very important issue. In the past, it's very time consuming to do the regular maintenance of cameras by people. When a camera breaks down, often critical images have been lost. In order to improve the efficiency of anomaly detection and reduce manpower. This paper proposes three methods for camera anomaly detection. Through the Video Management System, regularly capture camera snapshots and collect the time records of the video to analyze the connection and video status of the camera. When a disconnection or abnormal state is found, the system will automatically report for repair. In addition, through camera snapshots, the AI blur detection can be used to mark cameras with blurred or out-of-focus images.

並列關鍵字

surveillance camera ； anomaly detection ； deep learning ； blur detection

參考文獻

J. Zhou, S. Ntafos and B. Prabhakaran, "Fault Detection Framework for Video Surveillance Systems," in 2008 IEEE Fifth International Conference on Advanced Video and Signal Based Surveillance, pp. 219-226, New Mexico, 1-3 Sept. 2008

Y. -C. Zeng, C. -J. Yu and W. -T. Chang, "Fault detection of surveillance camera using fixed-point arithmetic operations," in 2014 IEEE International Conference on Consumer Electronics - Taiwan, 2014, pp. 87-88, Taipei, Taiwan,26-28 May. 2014

A. Sidnev, M. Barinova and S. Nosov, "Efficient Camera Tampering Detection with Automatic Parameter Calibration," in 2018 15th IEEE International Conference on Advanced Video and Signal Based Surveillance (AVSS), pp. 1-6, Auckland, New Zealand, 27-30 Nov. 2018

L. Dong, Y. Zhang, C. Wen and H. Wu, "Camera anomaly detection based on morphological analysis and deep learning," in 2016 IEEE International Conference on Digital Signal Processing (DSP), pp. 266-270, Beijing, China,16-18 Oct. 2016

蕭天佑,顏志平,林曾祥, "治安監錄攝影機畫面偏移異常自動偵測：以桃園市警用監視錄影系統為例, " 執法新知論衡,第 1 卷第17 期,中央警察大學,民國 110 年 6 月。