近來智慧型監控系統逐漸崛起，利用影像分析技術擷取有意義之事件資訊。但固定式攝影機之監控範圍有限並存在監視死角，有鑑於此，近年來興起的研究探討如何藉由高空載具，架設攝影機俯視地面攝影並分析影像，進行無死角之全面性監控。空中即時監控的困難點在於，民用無人飛機無法負載可轉動式攝影機，因此必須在地面目標上空進行盤旋，以側視方式攝影才能持續監控地面目標。然而利用無人飛行器上的感應器資訊，如GPS、電子羅盤、陀螺儀等，對地面目標點進行定位的誤差過大，無法成功的在目標上空盤旋。本論文提出一種地面目標定位技術，結合一可應用於無人飛行器即時影像之特徵點追蹤方法及無人飛行器空載感測器資訊，利用不同時間點影像中地面目標點的移動量及無人飛行器的移動距離資訊計算地面目標座標，取代傳流以誤差較大的無人機距地高度作為基礎的計算方式，可得到精確的地面目標座標。再傳送至無人飛機之自動導航系統，以地面目標點為中心進行盤旋，即可達成空中即時監控的目的。

The demand for video surveillance systems has been rising recently; a typical surveillance systemcaptures information of interesting event by video analytics technology. However the field of view of a fixedcamera in the video surveillance system is limited, and could not be expanded for a region-specific event.Many researches focused on the application of the video from unmanned aerial vehicles, like balloons orairplanes. The aerial video surveillance system is flexible for single or region-specific event. But right now thecommercial unmanned aerial vehicle is too light to carry a camera with Pan/Tilt function. To watch a target onthe ground continuously, an unmanned aerial vehicle with fixed-side-view camera must be in hover mode. Intheory, the sensors on the unmanned aerial vehicle (GPS/compass/gyro meter) could provide enoughinformation for target localization, but the precision is very poor due to sensor errors. Without precise locationof the target on the ground, it’s impossible to watch it continuously. The critical point of target localization isthe height of the unmanned aerial vehicle. We present a geolocation method for the unmanned aerial vehiclein this paper. It resolves the height of unmanned aerial vehicle by combining the feature point tracking method,and the sensor information of the unmanned aerial vehicle, to replace the height from GPS. With preciseheight and slope angle of camera, the target on the ground could be located. An unmanned aerial vehicle withauto-pilot function can hover on the target to watch it continuously.