匝道車輛辨識與計數

Abstract

隨著時代的進步，車輛已成為民眾生活中不可或缺的一部份，因此如何有效地交通控管便成為一個重要的課題。除了一般傳統車輛計數及辨識的儀器外，近年來以電腦視覺為基礎的車輛辨識系統也漸漸受到注目。大多數的車輛辨識系統將攝影機架設於道路上方，並且以車輛大小來分類車輛種類。這樣的車輛辨識系統不僅架設困難，且分類方法較不精細。 本研究中提供了一個以電腦視覺為基礎，不受天氣影響且架設方便的車輛辨識系統。有別於多數的車輛辨識系統，我們將攝影機架設於道路側面，以側拍的方式監測道路影像。攝影機架設於道路側面，不僅架設方便，無需負擔額外的架設費用外，利用車輛側面的形狀，我們更可以將車輛種類更細分為小客車、小卡車、大卡車、箱型車、貨車、載卡多、大客車，提供ㄧ個更精細的分類結果。另外本系統不受天氣影響，可用非常短的距離監測道路影像的特性，使得本系統能普便應用於各種車道環境。For many decades, vehicles have been an essential part of modern life. As the number of vehicles increased, transportation problems have become more and more serious. Tasks of vehicle counting and classification systems for vehicle classification based on computer vision have attracted much attention. In most such systems, a camera is set above a lane of traffic and classifies vehicles according to their sizes. Setting cameras above road is difficult and expensive, and classifying the vehicles according to their sizes is challenging.In this paper, we propose a system based on computer vision for vehicle classification and counting under different kinds of environments, and is also very convenient to install. Unlike most systems, we set up the camera beside a road. There is little additional cost to set up the camera. Based on the profile of a vehicle’s shape, we classify it one of many categories, such as sedan, flatbed truck, van, delivery truck, sport utility vehicle, and trailer truck. However, our system, which can work in all kinds of weather and in a confined space, should prove to be easy to adopt.