基於輪廓傅立葉頻譜之自動車牌辨識系統

隨著時代和電腦科技的進步，數位影像處理在各個領域的應用越趨廣泛，自動車牌辨識系統即為其一。本文的研究目的在探討基於不同數學分析方法在車牌辨識正確率的比較，其中包括三種方法：輪廓傅立葉描述子、質心輪廓距離及不變矩。首先本文透過影像前處理步驟，如：影像灰階化、邊緣偵測、影像形態學…等，擷取汽車前方影像中的車牌部份，再利用閥值最佳化的計算取得品質良好的二值影像，進行字元的分割。接著，利用邊緣萃取演算法取得所有字元的輪廓曲線，分別建置其傅立葉描述子頻譜、質心輪廓距離及不變矩的樣本數據檔。最後，從測試車牌中進行三種方法的正確率比較。實驗結果顯示：傅立葉描述子將影像的輪廓曲線視為週期函數，且不受影像旋轉、縮放、位移的影響，因此辨識結果最為優異，準確率高達90%以上。

關鍵字

車牌辨識；傅立葉描述子；質心輪廓距離；不變矩

並列摘要

As time progresses and technology improves, digital image processing has been widely used in various areas, where automatic license plate recognition (ALPR) system is one of the important applications. The purpose of this study is to compare the correctness of the ALPR system based on different mathematical analysis methods. There are three mathematical analysis methods used in this study including Fourier descriptor (FD),centroid-contour distance (CCD) and invariantmoment (IM). First, image pre-processing steps are used to obtain the license plate area in the car front image, such as RGB to grayscale conversion, edge detection, and image morphology. Next, a high quality binary image can be obtained by using threshold optimization calculation for segmenting characters from the image of license plate. Afterwards, we use the edge extraction algorithm to obtain the contours of all characters for building three sample date files of FD, CCD and IM, respectively. Finally, the correctness of some test license plate images using the three methods are compared in this study. The experimental results show that the correctness of the FD is the highest among these methods due to FD considers the contour of the image as the periodic functionand has the advantages of rotation, scaling, and translationinvariance. The accuracy rate of the FD method is up to 90% or more.

並列關鍵字

license plate recognition ； Fourier descriptor ； centroid-contour distance ； invariantmoment

參考文獻

[1] Intelligent Transportation System (ITS), http://www.iot.gov.tw/its/, 1999.

[2] CIE, Commission Internationale de l'Eclairage Proceedings. Cambridge, University Press, Cambridge, 1931.

[7] M. Sezgin and B. Sankur,“Survey over image thresholding techniques and quantitative performance evaluation,” Journal of Electronic Imaging, Vol. 13, No. 1, pp. 146–165, 2004.

[9] H. Kauppien,T. Sepanen, “An experiment comparison of autoregressive and Fourier based descriptors in 2D shape Classification,”IEEE Transactions on PAM I, Vol. 2, pp. 201-207, 1995.

[10] E. Persoonand K. S. Fu, “Shape discrimination using Fourier descriptors,”IEEE Transactions on PAM I, Vol. 8, pp. 388-397, 1986.

國際替代計量

基於輪廓傅立葉頻譜之自動車牌辨識系統

全文下載

主題瀏覽