智慧影像之應用 - 以城市安全為例

面對日益複雜的公共環境，中國的城市公共安全存在許多問題，很多單位都想在這領域貢獻能力，卻看到不少令人憂心的現象，例如：專業領域的研究能量過於分散、相關產業的研發人力不足、公司或研究單位各自為政及現行法律條文追不上領域的發展等問題。因此，如何做好城市安防是政府相關單位及產業界近年來高度重視的議題。城市公共安全最基本的建設是視訊監控系統，將這些龐大的影像與人工智慧相結合，就是現代城市公共安全平台最核心的智慧影像分析系統。這樣的系統架構不論在實作上，還是理論基礎上，都具有相當高的含金量與研究價值。本論文主要探討關於公共安全方面的管理方針、政策發展與實際案例，透過中國上海浦東新區城市智慧影像建設的現況，及智慧影像技術與公共安全之間的關係進行研究。從了解城市公共安全的需求、目的和發展趨勢開始，闡述視訊監控與智慧影像辨識系統於生活中的應用，接著提出城市公共安全系統的框架，並說明如何透過系統增進城市的公共安全防護，讓突發狀況能得到及時的處理。最後再討論如何持續改進系統框架，以期達到城市公共安全的永續發展。

關鍵字

城市安全；公共安全；視訊監控；人工智慧；智慧影像辨識

並列摘要

How to ensure security in cities is a vital issue both governments and industries pay close attention to. In China, numerous problems in public safety exist in the increasingly complicated environment and thus many units attempt to alleviate the worsening situation. However, some phenomena reveal that there is still much to be desired in this field. For instance, the research in the related field is only sporadic. Moreover, human resources are insufficient in the relevant research and development. Finally, the current regulations from the government fail to address the ever-worsening problems. Therefore, the present study aims to improve city security by analyzing data gained through surveillance. The basic infrastructure for public safety in cities is video surveillance system. Images captured in the video are processed through artificial intelligence, called the analysis system of intelligence image. Such combination is the most basic and essential system structure of public safety in cities. Thus the system structure is of great value both in practical application and academic research. This thesis primarily aims to investigate regulations, policy development, and case studies in public safety. The research target is Pudong New Area in Shanghai, China. The relationship among this area’s intelligence image infrastructure, intelligence imaging technique, and public safety is explored. The current study introduces the need, purpose, and development trend in public safety, and in turn elaborates on the application of video surveillance and intelligence image recognition in daily life, followed by the frame of public safety mechanism in cities. Furthermore, the researcher proposes methods to improve public safety in cities utilizing such a system for the purpose of timely emergency management. The final part of the thesis discusses ways to consistently improve the system frame in the hope of achieving sustainable development in public safety in cities.