應用紅外線熱影像分析技術於校園安全巡視系統之應用研究

Abstract

本研究提出一套利用紅外線熱影像分析技術結合機器人移動、雷射掃描與平台控制等功能所組成的校園安全巡視系統。 在科技發展快速之下，傳統需要由人力來進行巡邏的工作，逐漸被自動化電腦監視設備所取代，但在監視設備上很難確切掌握環境各個角落，因此，該如何有效地進行監視環境將是本研究的重點。 在系統整體動作主要分為環境建置階段與巡視運作階段，先找出環境中的固定熱源將其建置熱源資料庫，而在後續進行巡視過程中，持續監控熱源資料庫內的熱源。在熱源偵測的部分，對前方狀況做由下而上分段擷取作為尋找熱源動作，而熱源分析的部分包含影像處理、熱源特徵萃取、計算熱源中心和地圖座標等各種資訊。 最後，在本研究中採用一套系統性能指標，其中包含四個評估系統性能指標：準確性、有效性、敏感度和信賴度，並經由實驗測試結果顯示，不論是警示或警報的動作上都能達到85%以上。This study proposes a campus security patrol system that was using thermal image analysis technique, combined robot movement, laser scanning and platform control functions. With the rapid development of technology, the traditional of patrol work was needed to use the manpower, but it was gradually being replaced by automated of computer monitored equipment. However, the fixed monitoring equipment is difficult to master of the every corner of environment. Therefore, how to effectively monitor environment and find out the heat source will be focused on this study. In the overall system is divided into build the environment stage and patrol operational stage. First, find out the fixed heat source of environment and build the heat source database, and in the course of subsequent inspections, it will continue to monitor the heat source in the heat source database. In the detection of heat source, the environment captured images from the low to high position, it actions as looking for the heat source. In the analysis of heat source will get the various of information of heat source, and this action is contain image processing, extraction feature , calculate the center of heat source and tagging map coordinates, etc. Finally, this study uses a performance indicators to evaluation this system, it consists of four performance indicators: accuracy, validity, reliability and sensitivity, in the test results of the experimentshows, whether a warning or alert actions are can reach more than 85%.