Study on UAV Thermal Image Real-Time Tracking Based on Computer Vision Techniques

• Main Authors: HUANG,CONG-LIN, 黃琮淋
• Other Authors: CHEN,WEN-PING
• Format: Others
• Language: zh-TW
• Published: 2018
• Online Access: http://ndltd.ncl.edu.tw/handle/559ztw

碩士 === 國立高雄應用科技大學 === 電機工程系博碩士班 === 106 === 　　Because the related technology of DJI UAV develops rapidly and has many advantages, such as friendly price, high portability, simple operation, and easily assembling by users, many domestic and foreign manufacturers invest in UAV developing one after another, which leads UAV to become a very popular industry. But lots of UAV events has been occurred frequently and caused trouble and danger, such as UAVs went into aviation restricted areas and interfered in flights due to people's mistaken operation, collided famous buildings, and intruded military bases. In order to prevent such mistaken UAV events from happening, many kinds of anti-UAV system have been proposed successively for solving such problems. 　　Because UAV has small volume, hard-to-be-detected material, low sensitivity by general radar, and the system has highly building cost with limited performance. This thesis applies computerized visual technology to track the thermal image of UAV in real-time and expects to replace the expensive radar detecting way, and also uses video capture tool to capture the infrared thermal image in real-time, and then conducts image pre-processing. 　　Frame Difference analyzes the dynamic image tracking of the intruding UAV in real-time, calculating the centroid coordinate of the UAV and screen center displacement, transmitting the coordinate and displacement to the two-axis rotation platform, showing the UAV on screen during the tracking process. This study is conducted with DJI Mavic Pro. This system can achieve the dynamic image tracking of the intruding UAV in 100 meters in sunny day, and can achieve the dynamic image tracking of the intruding UAV in 100 meters in cloudy day.