本文以複雜環境背景的影像為考量對象,提出一套車牌自動定位的技術。並可結合車牌辨識技術,達到行進中車輛之車牌辨識系統,作為一般道路車輛之自動監控系統。 本文所提的方法可在影像中偵測任意方位、位置和大小的車牌,更甚者,車牌影像可以是在任意天候情況下取得的。我們的方法包括兩個主要步驟:移動物體偵測與移動式車牌定位。在移動物體偵測部份,利用影像差異法、投影分割改善與交集式投影輪廓粹取,準確地分割出我們所需的移動物體位置,以便偵測影像中各種類型的車輛,系統上移動物體偵測的正確率可達97.2%。 在車牌定位部份,系統上僅採用3x3區塊運算法的定位正確率達77.1%,加上車牌交越的限制條件後,系統之正確率改善為82.9%。再加上正確候選區和錯誤候選區的密度分析篩選,系統的定位正確率變為87.2%。初步車牌確認後,最後作投影分析的分割法,定位正確率可達91.4%。以上步驟證明我們提出的新方法與先前的方法綜合並行,不斷地提高系統的強健性與定位正確率,在實務系統上有不錯的結果。故此系統將可整合現有道路監視系統資源,作全台車輛之監控,以達失竊汽車協尋以及問題車之監控。
In this thesis, the license plate locating automatically system is researched and implemented on complex background image. The technology of moving vehicular license plate locating (MVLPL) and license plate recognition (LPR) for vehicular will integrate into the vehicular locating and monitoring (VLM) system. The VLM system is active and can be employed on the intelligent surveillance system for the running vehicle. In the moving object tracking system, four steps are employed. At first, the pre-processing on the input image is performed. Secondly, the vertical and horizontal projection of the image is accumulated. Thirdly, the image difference is completed. The simple algorithm is employed to detect the edge of moving objects obviously. Finally, using the mixing algorithm to demarcate boundaries of moving objects. The results of the experiments show that the performance of the moving objects tracking rate is 97.2%. In the license plate location system, because of the vehicular image have black and white characteristics focus on the license plate. We bring new algorithm like 3x3 blocks algorithm、crossing analysis、density analysis、project division and else. The results of the experiments show that the performance of the moving license plate locating rate is 91.4%. So the system could combine all of the monitoring systems on road and detect the license plate of running vehicular.