結合運用車身車牌車體顏色特徵之車輛識別方法

本論文提出一種運用於停車場之車身車牌車體顏色特徵結合一種嶄新的車輛識別方法，目的在於解決駕駛者進入停車場不需要在手持傳統停車票卡或者RFID感應卡以節省駕駛者時間。傳統無卡識別通常都採取車牌辨識，但因車牌汙垢及環境明亮影響有一定辨識瓶頸，所以本論文當車牌無法辨識時用了另外兩種車輛特徵車輛幾何特徵及車輛顏色特徵來輔助車牌特徵增加我們車輛辨識率。本論文則提出整合三種重要的車輛資訊：車牌特徵、車輛顏色、車輛幾何特徵，達成更有效、且計算量更低的車輛識別方法。在影像前處理前我們需要做車牌定位方面，我們利用垂直邊緣偵測找出車牌字體劇烈的變化來找出車牌位置;在車輛定位方面利用了水平投影及垂直投影找出車輛所在位置再做車輛特徵擷取。在車牌特徵方面，我們利用車牌字元切割及類神經網路樣板比對法(Neural Network Template Matching, NNTM)的技術及字型結構法來辨識車輛車牌及判斷新舊式車牌；並利用顏色色彩模型方法找出車輛顏色；在車輛幾何特徵方面，我們選用車輛寬度與車牌高度、寬度的比例關係來區別車型。在驗證方面，我們利用中原大學現行停車場的影像資訊來進行分析，在實驗結果上車牌定位辨識率為94.6%、在車輛定位上辨識率為95.5%、在車牌特徵辨識率為94.5%、在車輛幾何特徵辨識率為92.3%、在車輛顏色識別為89.3%及整體辨識率為97%，所以實驗結果證實所提方法的可行性與優越性比傳統的單一車牌辨識效果相當甚至更好。

關鍵字

車輛幾何特徵；車牌辨識；停車場識別入場方法；車輛顏色識別

並列摘要

This thesis applies the properties of geometric, color, and license plate feature by camera to identify a vehicle in a parking lot. Conventional ways to recognize license plate by camera did not work well when the plate is dirty or the environment in a parking lot is dark. This thesis integrates three important features of vehicle to recognize a vehicle. First feature is the recognition of center two characters of license plate. The way to identify new or old license plate in Taiwan is also studied. A neural network based template matching method is developed to recognize the license plate. Second image feature uses the widths and heights of vehicle and license plate to recognize the geometry features. Third image feature identifies the color of the car using HSV color space method. Also, the image pretreatment steps include locating the vehicle, license plate and segmentation of license plate characters. Experiments results show that the rates of proposed method to successfully locate and identify a license plate, vehicle and character segment is 94.6%, 95.5% and 91.6%. The rate to correctly identify the center characters of new or old plates is 94.5%. Besides, the rate to identify vehicle geometry features by proposed method is 92.3% and the rate to identify color of car is 89.3%. In license plate features identify to join vehicle geometry features and vehicle color features assist to identify. The total identification rate is 97%. Finally, this thesis also compares the proposed method with existing method in thesis experiment environment to show the improvement.

並列關鍵字

color features ； vehicle geometry features ； license plate features

參考文獻

[2] R. C. Gonzalez, E. Woods, “Digital Image Processing,” Addision-Wesley, 1992

[3] P. Comelli, P. Ferragina, M. N. Granieri, F. Stabile, “Optical recognition of motor vehicle license plates,” IEEE Trans on Vehicular Technology, vol. 44, no. 4, pp.790-799, 1995

[5] T. Naito, T. Tsukada, K. Yamada, K. Kozuka, S. Yamamoto, “Robust license-plate recognition method for passing vehicles under outside environment,” IEEE Trans on Vehicular Technology, vol. 49, no. 6, pp.2309-2319, 2000

[7] I. Anagnostopoulos, V. Loumos, E. Kayafas, “A License Plate-Recognition Algorithm for Intelligent Transportation System Applications,” IEEE Trans on Intelligent Transportation Systems, vol. 7, no. 3, pp. 377-392, 2006

[8] S. M. Youssef, S. B. AbdelRahman, “A smart access control using an efficient license plate location and recognition approach,” Expert Systems with Application, vol.34, pp. 256-265, 2008

國際替代計量

結合運用車身車牌車體顏色特徵之車輛識別方法

全文下載

主題瀏覽