智慧型停車場系統:車牌辨識,車輛導引,及裝置控制子系統

Intelligent Parking Lot System: License Plate Recognition, Vehicle Guidance, and Device Control Subsystems

張祥利(Shyang-Lih Chang)

指導教授 ： 陳世旺

國立臺灣師範大學/理學院/資訊工程研究所/博士(2009年)

摘要

本文提出一種新的停車場系統，稱為智慧型停車場系統，此系統結合影像處理、電腦視覺、感測、控制、通訊、網路、及先進的管理技術等，希望在現有的資源下做一有效的整合，以提供停車場管理者一個全自動、高效率的管理系統；並給予車主一個安全、便利、及人性化的服務。
本停車場系統含有七個子系統：車牌辨識系統、車輛導引系統、感測系統、控制系統、自動收費系統、網路及通訊系統、和中央管理系統。首先當車輛欲進入本停車場前，先以車牌辨識系統確認車輛的身份，並記錄其影像及進入的時間；接著由車輛導引系統，根據停車場內各停車區域車輛停放的情形，以方向指示器導引該車輛前往最佳停車位；當車輛停妥後，記錄其停放位置。而當車主欲前來取車離開時，可以車牌號碼、停放位置、或大約進場的時間中任何一種資訊，再配合影像確認車子，迅速繳費並取得車子；而在出口處亦設有車牌辨識系統，確認身份後決定是否開閘門放行。
此外，我們在停車場內各適當地點，分別安置有溫度、水位、煙霧、一氧化碳等感測元件，當停車場內有任何溫度、煙霧、一氧化碳濃度、水位異常時，可以適時的啟動場內自動控制系統，如消防設備、抽風、抽水馬達、防火、防水閘門、及警報系統。必要時亦可透過網路及通訊系統主動報警，及通知車主前來處理，以保障停車場內人員及車輛之安全。
在本研究中，我們除了提出整個停車場的規劃構想及理論架構外，主要著重在車牌辨識系統，車輛導引系統、及裝置控制系統的發展。亦提出一些實施的技術。其中發展演算法可自動規劃以最少的攝影機佈設，來達到監看整個停車場的目的；安全監視演算法自動察覺意外、治安、環境等的安全；車牌辨識演算法識別車輛的身份；車輛導引演算法可快速引導車輛至最佳停車位等。在硬體方面，我們嘗試將上述演算法及軟件中計算較為費時的部份，提出其硬體的架構，並以FPGA來實現，藉由硬體平行處理之能力，加速運算時間。此外對於停車場內之感測及控制等電路，則以單晶片微處理機來實現。

關鍵字

智慧型停車場系統 ； 車牌辨識 ； 車輛導引 ； 裝置控制 ； 彩色邊線偵測 ； 粒子濾波器

並列摘要

This study introduces a new parking lot management system called the Intelligent Parking Lot System (IPLS), which combines sensing, image processing, computer vision, controlling, communicating, networking, and advanced management technologies to ensure the smooth day-to-day operations of a parking lot. Using the Intelligent Parking Lot System, parking lot managers will be able to better manage and monitor their parking lot, as well as offer customers a safer and a more customer-focused environment.
The Intelligent Parking Lot System can be divided into seven sub-systems: a License Plate Recognition System (LPRS), Vehicle Guidance System (VGS), Security Surveillance System (SSS), Device Control System (DCS), Payment and Inquiry System (PIS), Network System (NS) and Central Management System (CMS). When a vehicle enters this parking lot, it is detected by the LPRS, which also takes a photo of the vehicle and records its entrance time. The Guidance System then leads the car to the most conveniently located vacant parking space through the direction indicators. After the car has been parked, the parking space will be updated in the system with its time stamp, the vehicle’s license plate number and another photo. Later on, when the owner wants to retrieve the vehicle, he or she can identify it in the system using information such as license plate number, parking space number, approximate entrance time or a photo of the car. After the customer has completed the payment process, the License Plate Recognition System then notifies the exit gate of the impending arrival of that particular vehicle, allowing the exit gate to successfully determine whether a particular vehicle should be allowed out or not.
Furthermore, sensors inside the parking lot are set up to measure temperature, water levels, smoke, and carbon dioxide concentration. If an abnormal state is detected, the automatic control system will be able to initiate contingency measures such as fire extinguishers, electric exhausts, pumping motors, fireproof and waterproof gates, as well as the alarm system. If necessary, the Network System will be able to inform the police, notify car owners to retrieve their vehicles, and ensure the general safety of the vehicles and the people inside the parking lot.
This research, in addition to presenting a complete parking lot system and an accompanying theoretical framework, also introduces some operational technologies and algorithms, which are listed below.
• License Plate Recognition algorithms can be used to identify vehicles.
• Development algorithms can be used to determine how to monitor the entire parking lot by using the least number of cameras.
• Car guidance algorithms can be used to lead the car to the most conveniently located parking space in a quick and efficient manner.
• Safety monitoring algorithms can be used to detect accidents and to monitor public safety.
Moreover, this study also explores developments in sensor controlling, automatic toll charging, Internet communication, and software for the central management system. A set of security delivery modes was also designed for data and information delivery among the sub-systems, with a focus on preventing unauthorized system intrusions, and also to facilitate an efficient information management system. We also tried to introduce the frameworks of the time-consuming algorithms and software mentioned above, into the hardware of the system. FPGA(Field Programmable Gate Array) was carried out to shorten the calculation time through the use of hardware parallel processing. Some Single-Chip micro-processors were also used to sense and control the various electric circuits of the parking lot system.

並列關鍵字

Intelligent Parking Lot System ； License Plate Recognition ； Vehicle Guidance ； Device Control ； Color Edge Detection ； Particle Filter

參考文獻

[Gon 93] R. C. Gonzalez and R. E. Woods, Digital Image Processing, New York, US: Addison-Wesley Publishing Co., pp. 197-201, 1993.

[Kel 94] J. M. Keller and R. Krishnapuram, “Fuzzy Decision Models in Computer Vision,” Fuzzy Sets, Neural Networks, and Soft Computing, R. R. Yager and L. A. Zadeh, Eds. New York: Van Nostrand, pp. 213–232, 1994.

[Als 04] F. E. Alsaqre, and Yuan Baozong, “Multiple Moving Objects Tracking for Video Surveillance Systems,” IEEE 2004 7th International Conference on Signal Processing, pp. 1301-1305 vol. 2, 2004.

[Bai 04] H. L. Bai and C. P. Liu, “A Hybrid License Plate Extraction Method Based on Edge Statistics and Morphology,” Proc. 17th Int. Conf. on Pattern Recognition, vol. 2, pp. 831-834, 23-26 Aug. 2004.

[Bla 02] J. Black, T. Ellis, and P. Rosin, “Multi View Image Surveillance and Tracking,” IEEE Workshop on Motion and Video Computing, pp. 169–174, 2002.