基於彩色影像之即時交通監控系統

Abstract

交通運輸在一個國家之中所扮演的腳色，猶如血液循環系統之於人體，其重要性不可言喻，而ㄧ套完整的交通監控系統，更有助於建立一個良善的交通運輸系統。對於先進國家而言，每年在擴充與維護良善的交通運輸系統上，可能都得花費龐大的經費來達到目的。然而，昂貴的經費與冗長的施工都會迫使擴建的速度嚴重落後。於是如何在現有的架構下，以目前已知的科技，如電腦、網路、通訊、控制、以及先進管理等技術，來提升交通運輸的效率、安全與舒適，便成為一個較為務實的目標。 本研究便是透過CCD所擷取的道路影像，即時的加以處裡、分析來達到即時的交通監控。在針對交通監控的影像處理中，背景圖往往是最重要的依據，完整的背景圖對於後續的分析有著深遠的影響。我們針對於一連串的彩色，大小為320*240的靜態道路影像，利用統計分析與機率選取的模式，提出了一個可以快速獲得背景圖的方法，在獲得背景後，就可以藉由背景圖與後續影像來做交通參數的分析與判斷，以達到交通監控的目的。

The importance of the transportation system to a country can be just like that of the blood circulation system to the human body. Many countries have spent a great amount of annual budget for maintaining, improving, and enhancing their transportation systems. It was common in the past that hardware resources (e.g., roads and vehicles) were introduced in order to meet the rapidly increasing requests on transportation. However, the former almost always hardly catches up with the latter. Moreover, introducing hardware equipments could be inadequate for the areas with limited sizes and crowded people. Therefore, under present conditions how to increase the efficacy, safety, and comfort of transportation systems with the help of currently available high technologies, such as computers, networks, communication, controls, and advance managements may become more practicable. In this study, we propose a moving object segmentation approach based on an efficient and real-time color background extraction. We use the statistical algorithm to obtain the color background and moving objects information efficiently. Then we perform the run-length algorithm to compensate the segmented objects and update the background real-time. The variation of the illuminative condition is also considered in our work. Several experimental results are illustrated to show the advantage of the proposed work on processing time and the segmented result.