發(fā)布時(shí)間：2018-06-24 02:13

  本文選題：運(yùn)動(dòng)目標(biāo)檢測(cè) + 車牌識(shí)別　； 參考：《西安科技大學(xué)》2016年碩士論文

【摘要】：近年來(lái),科學(xué)技術(shù)呈現(xiàn)出跳躍式的發(fā)展,技術(shù)改革日新月異,尤其是計(jì)算機(jī)和計(jì)算機(jī)視覺(jué)技術(shù)的發(fā)展,成為當(dāng)今社會(huì)研究中最主要的熱點(diǎn)。利用高新技術(shù)取代人類活動(dòng),解放勞動(dòng)力,提高生產(chǎn)效率和精度一直是人們致力研究的問(wèn)題,而人類活動(dòng)(生產(chǎn)、生活)中信息的交流中,80%左右的信息是視覺(jué)信息。車輛通行智能門禁系統(tǒng)是智慧城市的重要內(nèi)容之一,融合了智能檢測(cè)、車牌識(shí)別以及自動(dòng)門禁控制系統(tǒng),是數(shù)字圖像處理、計(jì)算機(jī)技術(shù)以及自動(dòng)化技術(shù)在實(shí)際生活中的一個(gè)重要應(yīng)用。針對(duì)社區(qū)車輛門禁智能管理系統(tǒng)設(shè)計(jì)工作的研究,本文主要對(duì)其中基于圖像處理技術(shù)的智能識(shí)別與檢測(cè)部分展開研究,研究工作可以分為成4個(gè)部分:車輛通行事件檢測(cè)、車牌定位、字符分割以及字符識(shí)別。首先,車輛通行檢測(cè)研究中,我們采用了基于運(yùn)動(dòng)目標(biāo)檢測(cè)的思想,通過(guò)背景差分方法對(duì)檢測(cè)區(qū)域內(nèi)的目標(biāo)進(jìn)行檢測(cè),利用車輛相對(duì)于行人、自行車等干擾目標(biāo)在形狀和尺寸等方面的本質(zhì)差異進(jìn)行判別得到車輛通行事件的檢測(cè)結(jié)果。在背景建模過(guò)程中,我們采用了混合高斯模型來(lái)建立模型,進(jìn)行目標(biāo)的檢測(cè)。其次,對(duì)于車牌定位技術(shù)的研究,車牌定位方面,由于門禁系統(tǒng)應(yīng)用場(chǎng)景的簡(jiǎn)單性,在算法設(shè)計(jì)中更加具有針對(duì)性,更加強(qiáng)調(diào)車牌的色彩,同時(shí)考慮到黑底白字和白底黑字車牌的存在,結(jié)合紋理、邊緣梯度信息,利用灰度形態(tài)學(xué)處理技術(shù),設(shè)計(jì)定位算法。在車牌識(shí)別方面,由于本文中門禁系統(tǒng)應(yīng)用場(chǎng)景簡(jiǎn)單,車牌圖像樣本狀況較好,對(duì)車牌矯正和切割技術(shù)要求較低,因此本文中分別采用線性回歸方法和垂直投影技術(shù)進(jìn)行車牌矯正和字符切割。然后對(duì)于字符識(shí)別,我們采用現(xiàn)在最先進(jìn)的深度學(xué)習(xí)網(wǎng)絡(luò)進(jìn)行字符識(shí)別,首先通過(guò)深度信度網(wǎng)絡(luò)模型,建立生成式概率模型,其次將模型展開,構(gòu)建卷積神經(jīng)網(wǎng)絡(luò)模型,進(jìn)行字符的分類識(shí)別。最后,基于matlab仿真平臺(tái)對(duì)系統(tǒng)進(jìn)行了界面開發(fā)設(shè)計(jì),將智能門禁系統(tǒng)統(tǒng)一到同一平臺(tái),實(shí)現(xiàn)人機(jī)交互,便于算法研究和同一的管理。
[Abstract]:In recent years, science and technology has been developing by leaps and bounds, and technological reform is changing with each passing day, especially the development of computer and computer vision technology, which has become the most important hot spot in the social research nowadays. Using high and new technology to replace human activities, liberate labor force, improve production efficiency and precision has been a problem that people have been working hard to study, and in the exchange of information in human activities (production, life), about 80% of the information is visual information. The intelligent access control system is one of the important contents of intelligent city. It combines intelligent detection, license plate recognition and automatic access control system. It is a digital image processing system. Computer technology and automation technology in real life an important application. In view of the research on the design of community vehicle access control intelligent management system, this paper mainly studies the intelligent recognition and detection based on image processing technology. The research work can be divided into four parts: vehicle traffic event detection, License plate location, character segmentation and character recognition. First of all, in the research of vehicle traffic detection, we adopt the idea of moving target detection, using the background differential method to detect the target in the area, using the vehicle relative to the pedestrian, The detection results of vehicle traffic events are obtained by discriminating the essential differences in shape and size of interfering targets such as bicycles. In the process of background modeling, we use the mixed Gao Si model to build the model and detect the target. Secondly, for the research of license plate location technology and license plate location, because of the simplicity of the application scene of the access control system, it has more pertinence in the algorithm design and more emphasis on the color of the license plate. At the same time, considering the existence of white characters and black characters on black background, combined with texture and edge gradient information, the location algorithm was designed using grayscale morphology processing technology. In the area of license plate recognition, because of the simple application scene of the entrance control system in this paper, the status of license plate image samples is good, and the requirements for license plate correction and cutting technology are lower. Therefore, linear regression method and vertical projection technique are used to correct license plate and cut characters. Then for character recognition, we use the most advanced depth learning network for character recognition. Firstly, through the depth reliability network model, we establish the generated probability model, and then expand the model to construct the convolution neural network model. Classification and recognition of characters. Finally, the interface of the system is designed based on the matlab simulation platform, and the intelligent access control system is unified into the same platform to realize human-computer interaction, which is convenient for the algorithm research and the management of the same system.
【學(xué)位授予單位】：西安科技大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2016
【分類號(hào)】：TP391.4;TP311.52

【參考文獻(xiàn)】

相關(guān)期刊論文 前9條

1 郭榮艷;胡雪惠;;BP神經(jīng)網(wǎng)絡(luò)在車牌字符識(shí)別中的應(yīng)用研究[J];計(jì)算機(jī)仿真;2010年09期

2 許志遠(yuǎn);柳曉鳴;;基于雙線性插值動(dòng)態(tài)直方圖均衡化的霧天圖像增強(qiáng)算法[J];大連海事大學(xué)學(xué)報(bào);2010年03期

3 張艷陽(yáng);顧明;;基于AdaBoost分類器的車牌字符識(shí)別算法研究[J];計(jì)算機(jī)應(yīng)用研究;2006年05期

4 李貴俊,劉正熙,游志勝,莊永;一種基于色差和彩色歸一化的車身顏色識(shí)別算法[J];計(jì)算機(jī)應(yīng)用;2004年09期

5 包明,路小波;基于Hough變換的車牌傾斜檢測(cè)算法[J];交通與計(jì)算機(jī);2004年02期

6 吳大勇,魏平,侯朝楨,劉永信;一種車牌圖像中的字符快速分割與識(shí)別方法[J];計(jì)算機(jī)工程與應(yīng)用;2002年03期

7 王廣宇;車輛牌照識(shí)別系統(tǒng)綜述[J];鄭州輕工業(yè)學(xué)院學(xué)報(bào);2001年02期

8 魏武;黃心漢;張起森;王敏;王明俊;;基于模板匹配和神經(jīng)網(wǎng)絡(luò)的車牌字符識(shí)別方法[J];模式識(shí)別與人工智能;2001年01期

9 王昱,趙正校,楊碩;基于直線邊緣識(shí)別的圖象區(qū)域定位算法[J];計(jì)算機(jī)工程;1999年09期

相關(guān)碩士學(xué)位論文 前4條

1 所瑪;基于小波變換和神經(jīng)網(wǎng)絡(luò)的車牌識(shí)別系統(tǒng)研究[D];黑龍江大學(xué);2009年

2 許利顯;車牌識(shí)別系統(tǒng)中關(guān)鍵技術(shù)的研究[D];西北大學(xué);2008年

3 王笛;基于數(shù)字圖像處理和機(jī)器學(xué)習(xí)的車牌識(shí)別（LPR）研究[D];重慶大學(xué);2008年

4 黎婷婷;車牌識(shí)別圖像處理算法的研究與實(shí)現(xiàn)[D];武漢理工大學(xué);2007年

，

本文編號(hào)：2059520