影像辨識在停車場監控之應用與設計

Abstract

本篇論文利用影像處理的技術結合資料庫查詢，發展一套停車場的導引系統。由於傳統的停車場只能告知駕駛目前停車場尚有幾個空位，對於停車位的位置資訊則是完全無法提供訊息。所以希望透過影像處理的方式來得知車輛的位置，進而提供導航資訊給駕駛者，帶領車輛到選定之停車位。系統的主要功能為多物體追蹤以及路徑導航，在追蹤方面，我們利用影像的連續性來預測物體可能出現的位置，並且加入了碰撞偵測的功能。此外，我們設計了一套多部攝影機合作追蹤的機制，使得系統能夠監控大範圍的區域。在導航方面，透過查詢資料庫的路徑導航資料，可以在每個路口之前根據車輛的位置、行進方向及目的地得知車輛應該行進的方向。

In this thesis, we develop a guiding system for parking lot based on image processing and database. In traditional parking lot, drivers know the count of the empty parking spaces, but have no idea where the available parking space is. Consequently, we hope to get vehicle’s position in image processing way, further provide guiding information to drivers then guide vehicles to their chosen parking space. The main functions of the system are multiple-object tracking and guidance. In the tracking work, we predict the object’s position by image’s continuity, detect the occurrence of bump condition as well as develop wide area monitoring with the aid of multiple-camera cooperation mechanism. In guiding work, we are able to obtain the guide direction by searching guiding data in database according to vehicle’s position, moving direction and destination.