Smart City: Smart Cloud Transport Vehicles Video Surveillance Video Assessment Mechanism Analysis and Discourse

• Main Authors: Yao-Wen Kan, 甘耀文
• Other Authors: You-Shyang Chen
• Format: Others
• Language: zh-TW
• Published: 2015
• Online Access: http://ndltd.ncl.edu.tw/handle/8um8u6

碩士 === 華夏科技大學 === 資訊科技與管理研究所碩士在職專班 === 104 === Transportation is what drives modern civilization forward. As the number of motorized vehicles increases, so does that of traffic accidents involving these vehicles in Taiwan every year, including traffic violations, criminal behavior patterns, traffic-related incidents and DUIs. These incidents do no decrease despite the efforts of the government to prevent them. According to the latest numbers of National Police Agency, Ministry of the Interior, 3,129 were killed in accidents involving motorized vehicles in 2013 alone, and this number was a staggering 2,103 in 2014 up to October. The financial expenses, manpower and medical supports were skyrocketing. When it comes to the major cause, it can be the poor images recorded by the surveillance cameras installed by local governments and administrations, inefficient equipment provided, lack of effective management, and low camera definition resulting in difficulties in providing evidence for local authorities either for justification or improvement. As a comparison, the number of traffic-related and criminal cases of unknown causes is also significant. As an effort to deal with the poor quality of surveillance equipment and inefficient management, this study was intended to investigate the possibility of establishing an “intelligent cloud-based transportation vehicle image tracking and evaluation mechanism” and image tracking analysis based on the concept of intelligent monitoring that meets the practical demands of a modern city. The idea was centered on dynamic identification of license plate. The image of a license plate was digitally converted and transmitted to an integrated cloud-based database management system for classification, management, analysis and application, as to improve the capability of digital monitoring of a metropolis.  The existing license plate identification algorithm was studied and developed into an improved combination of active contour algorithm and improved differential algorithm. Based on this combination, a smart monitoring and dynamic license identification algorithm was realized specifically for an intelligent city. For verification, the 2nd place of the top ten locations of most accidents in New Taipei City, which is Zhongzheng Road in Zhonghe District, was chosen as the proving ground. For example, a 20-minute footage captured the image of 711 vehicles and 95.63% of license plates were scanned with 93.80% of correct identification. The experiment results indicated that the active license plate identification algorithm proposed achieved accurate multi-license locating and dividing techniques, license plate image noise removal and stable-unstable light source processing in complicated environment conditions and ensured accurate identification of license plates with the help of high-definition cameras. Also, the integration of cloud-based panning and management mechanism contributed to the achievement of tracking vehicles that escaped civil and criminal cases, clarification of responsibility in traffic accidents, identification of causes, and traffic flow statistics.