發(fā)布時(shí)間：2018-06-25 15:06

  本文選題：車(chē)牌定位 + 多特征與多方法�。� 參考：《西安電子科技大學(xué)》2014年碩士論文

【摘要】：伴隨著科學(xué)技術(shù)的飛速發(fā)展、日常生活質(zhì)量的全面提高，，汽車(chē)的數(shù)量在各個(gè)國(guó)家的總體數(shù)量都出現(xiàn)出了快速的增長(zhǎng)。目前我國(guó)的汽車(chē)數(shù)量已經(jīng)增加到一億兩千萬(wàn)輛，是世界汽車(chē)總數(shù)量的12%，已經(jīng)成為了世界上第二大的汽車(chē)國(guó)家，并且其數(shù)量仍然在快速的增長(zhǎng)當(dāng)中。當(dāng)然，隨著汽車(chē)數(shù)量的不斷增多，馬路的不斷加寬加長(zhǎng)，對(duì)這些汽車(chē)的管理也就是現(xiàn)在的交通問(wèn)題越來(lái)越被人們所關(guān)注，從而使其成為了一門(mén)很重要的學(xué)科來(lái)研究。在這門(mén)學(xué)科中主要研究的問(wèn)題又依賴于對(duì)車(chē)輛的識(shí)別，車(chē)輛識(shí)別也就是車(chē)牌的識(shí)別。 一般車(chē)牌識(shí)別的步驟主要有三部分內(nèi)容：車(chē)牌定位，車(chē)牌字符分割，車(chē)牌字符識(shí)別。這三個(gè)部分組成了一個(gè)完整的車(chē)牌識(shí)別的過(guò)程。由于精力有限，本文只對(duì)三個(gè)步驟中的車(chē)牌定位和字符分割進(jìn)行相應(yīng)的仿真研究。 本文用到的定位方法是根據(jù)車(chē)牌的字符本身的橫向和縱向掃面特征、車(chē)牌的字符和底板的色彩差別特征，再結(jié)合其他的特征，通過(guò)特定算法的篩選過(guò)程來(lái)確定車(chē)牌位置并提取出車(chē)牌圖像。主要算法是通過(guò)比較決定使用Prewitt算子邊緣檢測(cè)提取車(chē)牌的邊緣信息，根據(jù)車(chē)牌的垂直方向掃面特征和周邊色彩區(qū)別特點(diǎn)去除一定的干擾邊緣信息，連接邊緣點(diǎn)，標(biāo)記連通域，根據(jù)車(chē)牌的形狀特征并采用Adaboost方法繼續(xù)去除干擾信息凸顯車(chē)牌信息以分離出車(chē)牌圖像，對(duì)分離出來(lái)的車(chē)牌圖像傾斜校正，去除上下邊界與左右邊界得到精確的車(chē)牌字符區(qū)域。經(jīng)過(guò)仿真分析，本文算法能夠很好的定位出車(chē)牌圖像，對(duì)單車(chē)牌圖像和多車(chē)牌圖像都適用，定位成功率高，通用性好。 本文的另外一部分是車(chē)牌分割，這部分內(nèi)容第一步先分析了幾種典型的字符分割的方法，使用了基于車(chē)牌垂直方向投影法與先驗(yàn)知識(shí)作為模板匹配結(jié)合的車(chē)牌字符分割方法來(lái)將前面經(jīng)過(guò)精確定位的車(chē)牌字符圖片分割成單獨(dú)的字符。
[Abstract]:With the rapid development of science and technology, the overall improvement of the quality of daily life, the number of cars has increased rapidly in the total number of countries. At present, the number of cars in our country has increased to one hundred and twenty million vehicles, 12% of the total number of cars in the world, and has become the second largest automobile country in the world, and the number of cars in the world has become the second largest car country in the world. The number is still growing rapidly. Of course, with the increasing number of cars and the widening of the road, the management of these cars is becoming more and more concerned about the current traffic problems, which makes it a very important subject to study. Vehicle recognition and vehicle recognition are license plate recognition.
There are three main steps in the general license plate recognition: license plate location, license plate character segmentation, and license plate character recognition. These three parts make up a complete process of license plate recognition. Because of the limited energy, this paper only carries out the corresponding simulation research on the license plate location and character segmentation in the three steps.
The location method used in this paper is based on the characteristics of the transverse and longitudinal sweep of the character of the license plate, the characteristics of the color difference between the character of the license plate and the floor, and then combining the other features, the location of the license plate is determined by the selection process of the specific algorithm and the license plate image is extracted. The main calculation method is to determine the edge detection using the Prewitt operator by comparison. The edge information of the license plate is extracted, the interference edge information is removed according to the characteristics of the vertical sweep surface and the peripheral color, the edge points are connected and the connected domain is tagged. According to the shape feature of the license plate and the Adaboost method, the vehicle license information can be removed to separate the license plate image. The license plate image is inclined to correct, and the accurate license plate character area is obtained by removing the upper and lower boundary and the left and right boundaries. After simulation analysis, this algorithm can locate the license plate image well. It is suitable for both the single license plate image and the multiple license plate image, and the success rate of the location is high and the versatility is good.
The other part of this paper is the license plate segmentation. In this part, the first step is to analyze several typical character segmentation methods, and use the license plate character segmentation method based on the vertical direction projection method and the prior knowledge as template matching to divide the previously accurately located license plate character pictures into separate characters.
【學(xué)位授予單位】：西安電子科技大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2014
【分類號(hào)】：U495;TP391.41

【參考文獻(xiàn)】

中國(guó)期刊全文數(shù)據(jù)庫(kù) 前10條

1 韓麗萍;藥春暉;張文格;尹王保;;基于小波變換與形態(tài)學(xué)的車(chē)牌定位方法[J];測(cè)試技術(shù)學(xué)報(bào);2006年01期

2 李波,曾致遠(yuǎn),付祥勝;基于數(shù)學(xué)形態(tài)學(xué)和邊緣特征的車(chē)牌定位算法[J];電視技術(shù);2005年07期

3 王杰;李洪興;王加銀;苗志宏;;一種圖像快速線性插值的實(shí)現(xiàn)方案與分析[J];電子學(xué)報(bào);2009年07期

4 琚旭;王浩;姚宏亮;;基于Boosting的支持向量機(jī)組合分類器[J];合肥工業(yè)大學(xué)學(xué)報(bào)(自然科學(xué)版);2006年10期

5 陳寅鵬,丁曉青;復(fù)雜車(chē)輛圖像中的車(chē)牌定位與字符分割方法[J];紅外與激光工程;2004年01期

6 陳黎,黃心漢,王敏,李煒;基于聚類分析的車(chē)牌字符分割方法[J];計(jì)算機(jī)工程與應(yīng)用;2002年06期

7 賈曉丹;李文舉;王海姣;;一種新的基于Radon變換的車(chē)牌傾斜校正方法[J];計(jì)算機(jī)工程與應(yīng)用;2008年03期

8 李亞軍;劉曉霞;陳平;;改進(jìn)的AdaBoost算法與SVM的組合分類器[J];計(jì)算機(jī)工程與應(yīng)用;2008年32期

9 張順利;李衛(wèi)斌;吉軍;;基于投影的文檔圖像傾斜校正方法[J];計(jì)算機(jī)工程與應(yīng)用;2010年03期

10 王興玲;;最大類間方差車(chē)牌字符分割的模板匹配算法[J];計(jì)算機(jī)工程;2006年19期

中國(guó)博士學(xué)位論文全文數(shù)據(jù)庫(kù) 前1條

1 高鵬毅;BP神經(jīng)網(wǎng)絡(luò)分類器優(yōu)化技術(shù)研究[D];華中科技大學(xué);2012年

本文編號(hào)：2066424