三維空間多攝影機視覺監控系統

Abstract

近年來，由於數位影像處理技術的進步，各種相關應用也隨之發展，其中之一即為視覺監控系統(Visual-based surveillance system)。在本篇論文中，我們利用一般監控用攝影機對場景進行監控，將目前影像與背景影像配合向量線性相依(Linear dependence of vector)原則比對得到影像內新植入的目標物，並雜訊、陰影等造成的影響進行清除處理。而後藉由相機校正的程序得相機的內部參數K及外部參數R、T。根據這些相機參數進一步推算目標物的三維座標資訊，並利用這個三維座標指揮另一部校正好的相機進行目標物鎖定與特寫。將來進一步發展，可以進行三維目標物的追蹤及事件分析的工作。

Recently, digital image processing technology has made great progress; there are tons of applications about it. One major application is the digital visual surveillance system. In this thesis, we want to monitor the scene with commercial cameras, detect moving object in the image by comparing the current image and background image, and reduce the effects of noise or shadow as much as possible. We calibrate our cameras, to obtain the intrinsic parameter K and extrinsic parameters R, T. With these parameters we may compute the 3D world coordinate of the object and control another calibrated camera to take a close-up picture of that object later. In the future, we are going to monitor a sequence of images to track the moving object and analyze the history of object motion for activity analysis.