夜間自走車安全巡邏

Abstract

本論文研究主要是建構一種強調夜間監控的自走車裝備系統，不僅只日間使用。在實作中使用備有近紅外線照明的監控攝影機，再將其裝置在自走車上，它可依使用者的需求自由行走，以避免傳統監控的攝影死角，可預設定時間監控巡邏外，並透過無線傳輸的方式，使其具備遠端操控的功能，也增進系統的便利性。此種有別於傳統的監控系統既實用又簡便，使用者只需在遠端先針對要監控的物品，去設定自走車移動的路徑，並圈選物品的範圍大小，並依據物品外觀變化及灰階分佈情形，去判斷物品的安全與否。因此，本論文採用Snake Algorithm ( Active Contour Model ) 先去取得監控物品輪廓及灰階直方圖，再以Hausdorff Distance及Histogram Intersection為判斷方式，最後經由實驗結果證實其可行性及正確性。

The main purpose of this thesis is to build a mobile robot system for the nighttime surveillance, which can also be used in the daytime. The system uses an IR camera and wireless device to make the mobile robot move unrestrictedly and can navigate at night at scheduled time. Through the wireless device, the remote control makes the system more convenient to use. Different from traditional surveillance system, users only need to set up the route for the mobile robot once by leading the robot toward the monitored object. For an object selected to be monitored, the system will record its contour and gray histogram as object features for object matching during the security patrolling. In this thesis, we estimate the contour of the object by Snake Algorithm ( Active Contour Model ), and with the gray histogram,do the judgment by using Hausdorff Distance and Histogram Intersection. The results of the experiment show the system is feasible can work satisfactorily.