靜態影像車位狀態偵測; Detection of Parking Space Status by Using a Static Image

NCUIR > College of Electrical Engineering & Computer Science > Executive Master of Computer Science and Information Engineering > Electronic Thesis & Dissertation > Item 987654321/8776

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/8776

Title:	靜態影像車位狀態偵測;Detection of Parking Space Status by Using a Static Image
Authors:	蔡偉凌;Wei-Ling Tsai
Contributors:	資訊工程學系碩士在職專班
Keywords:	群集分析;影像處理;停車格監控;k-means cluster analysis;image processing;parking space detection
Date:	2009-07-06
Issue Date:	2009-09-22 11:34:36 (UTC+8)
Publisher:	國立中央大學圖書館
Abstract:	停車格狀態偵測是停車場監控管理系統的重要技術之一。在本篇論文中我們提出一個利用單個監視器獲取的單張影像來評估目前停車格狀況的演算法。通常一般最簡單的辨識方法為在每個停車格上面架設個別的感測器，但再加上後續的維護，這將會是一筆不小的費用。因為目前一般的停車場內都已經有架設監視器，我們可利用這些監視器拿來作監控系統利用，因此可另外省下購買與架設的費用。本篇論文以電腦視覺為基礎，利用影像處理的技術發展出一種新穎的停車格狀態偵測演算法。演算法內容為結合停車格的特徵萃取與k-means分群法來判別目前停車格中的停車狀態。最後我們將所提出的演算法實際在一個戶外停車場作驗證，總共測試了五天影像中涵蓋各種時間及天氣，如早中晚、陰天、晴天及起霧等各種天氣狀態。測試結果證實了我們所提出演算法的精確性，正確率高達98%。 Parking cell detection is one of the key technologies in parking lot monitoring and management system. In this thesis, we propose an algorithm to estimate the occupancy of a parking cell using a single image captured by a camera. Usually, car-parks are already equipped with CCTV-cameras for surveillance purpose which can be served for automatic detection systems as well. Our system is targeted on whether a parking cell is occupied or not. Exact solutions like individual sensors are too costly. In this thesis, we propose a vision-based system that use computer vision techniques for detecting the occupation status by using the static image captured by a single camera. Our algorithm uses the combination of parking cell features extraction and k-means clustering to discriminate occupancy status of parking spaces. The proposed method was tested on an actual outdoor parking lot for a period of 5 days with different weather conditions from sunrise to sunset. The results confirm the feasibility of the proposed method with the accuracy rate being over 98%.
Appears in Collections:	[Executive Master of Computer Science and Information Engineering] Electronic Thesis & Dissertation

Files in This Item:

File	Size	Format

社群 sharing

Loading...