基於DaVinci 平台的車牌定位系統

License Plate Location System Based on DaVinci Platform

林建良(Jian-Liang Lin)

指導教授 ： 范育成

國立臺北科技大學/電資學院/電子電腦與通訊產業研發碩士專班/碩士(2010年)

摘要

在智慧型運輸系統的應用中，車牌辨識系統是相當關鍵的技術，例如停車場管理，車輛違規以及查緝贓車等。車牌辨識演算法一般是由以下三個處理步驟所組成：(1)車牌定位(License plate location，LPL)；(2)車牌字元分割；(3)字元辨識。其最重要的關鍵是車牌定位，本論文針對車牌定位提出一個動態比對演算法及電路設計，本篇論文提出的方法主要可分為前處理和後置處理兩部份，前處理的部份是影像輸入後進行Gray、Resizing、直方圖均化等處理，主要是將影像擷取後轉換成灰階資訊，並針對太陽光的影響作處理，其中對太陽光的處理我們搭配修改後的直方圖均化使得車牌和背景的差異變大，本實驗結果得到不錯的效果；後置處理則採用Sobel、Dilation、Labeling找出最大的標籤區域以及作區域過濾找到最後的遮罩，接著使用投影法找到的相對座標與原始圖像對映，後置處理主要是針對車牌特性作的處理，其中投影法我們使用動態的方式找尋符合車牌的比例，不會因為車牌太大或太小，而使得找到的車牌不完全。
　　本論文的貢獻在於單張影像處理時，能針對太陽光的影響作修正以及能動態找尋符合車牌的比例，在執行時間的結果為0.062秒，車牌定位率為97.80%，並於DaVinci平台下完成即時車牌定位系統，此論文所提出的車牌定位方法，具有很好的實用性。

關鍵字

車牌定位 ； 連通法 ； DaVinci DM6446

並列摘要

License plate recognition (LPR) system is crucial to technology of intelligent transportation system application, such as car park management, vehicle violations, as well as the seizing stolen cars and so on. LPR algorithms are generally composed of the following three processing steps: 1) license plate location (LPL); 2) segmentation of the plate characters; and 3) recognition of the plate character. LPL is most important step of LPR system. We propose a dynamic location for the license plate matching algorithm and circuit design. The proposed method can be divided into pre-processing and post-processing parts. Pre-processing is using gray, resizing, histogram equalization algorithms and reduces the sunlight effects. We use modified histogram equalization algorithm to increase the difference between the license plate and background and obtain good results. Post-processing is using sobel, dilation, labeling algorithms to find the largest region and the final mask. It uses the projection method to find the relative coordinates, finally mapping with original image. The main target of post-processing is handling characteristics for the plates. We use the dynamic projection method to find consistent with the ratio of plate.
　　In this thesis, we overcome the sunlight effects and perform dynamically search of license plate location system. At the same time, the proposed scheme increases the location ratio. The execution time is 0.062 seconds; license plate location rate is 97.80%. We adopt DaVinci platform to achieve the license plate location system. The proposed approach provides practical application and well solution for license plate location system.

並列關鍵字

License Plate Location (LPL) ； Connected Component Labeling (CCL) ； DaVinci DM6446

參考文獻

[2] B. Hongliang and L. Changping, “A hybrid license plate extraction method based on edge statistics and morphology,” in Proc. ICPR, pp. 831–834, 2004.

[3] D. Zheng, Y. Zhao, and J. Wang, “An efficient method of license plate location,” Pattern Recognit. Lett., vol. 26, no. 15, pp. 2431–2438, Nov. 2005.

[4] S. Wang and H. Lee, “Detection and recognition of license plate characters with different appearences,” in Proc. Conf. Intell. Transp. Syst., vol. 2, pp. 979–984, 2003.

[6] N. Bellas, S. M. Chai, M. Dwyer, and D. Linzmeier, “FPGA implementation of a license plate recognition SoC using automatically generated streaming accelerators,” in Proc. 20th IPDPS, Nice, France, pp. 8–15, Apr. 2006.

[8] F. Yang and Z. Ma, “Vehicle license plate location based on histogramming and mathematical morphology,” in Proc. 4th IEEE Workshop Autom. Identification Advanced Technol., pp. 89–94, 2005.