發(fā)布時間：2018-03-30 19:44

  本文選題：智能交通　切入點：數(shù)字圖像處理　出處：《重慶交通大學(xué)》2015年碩士論文

【摘要】：智能交通系統(tǒng)(ITS)以交通運輸、服務(wù)控制和車輛制造為研究對象,以解決交通擁堵、增強車輛安全性、提高汽車智能化為研究目的。道路交通標(biāo)志識別(TSR)是智能交通系統(tǒng)中對于服務(wù)控制研究的重要組成部分,其研究主要需要電子信息、視頻處理、數(shù)字圖像處理、人工智能和模式識別等技術(shù)手段。目前,交通限速標(biāo)志檢測識別方法普遍存在識別正確率不理想,識別速度不夠快,對于殘損交通限速標(biāo)志難以識別等問題。為此,本文選擇對交通限速標(biāo)志的檢測識別方法進行研究,采用廣義結(jié)構(gòu)元進行標(biāo)志檢測,設(shè)計新的模板匹配識別算法,擬解決識別率不高、識別速度慢和對于殘損交通限速標(biāo)志的識別難等問題。主要研究內(nèi)容如下:(1)圖像預(yù)處理。比較RGB和HSI空間對于交通限速標(biāo)志檢測的優(yōu)缺點,選用HSI空間對交通限速標(biāo)志進行處理。在此空間中對圖像進行直方圖增強,維納濾波,幾何變換等預(yù)處理。(2)廣義結(jié)構(gòu)元濾波。引入廣義結(jié)構(gòu)元思想進入交通限速標(biāo)志檢測識別中,構(gòu)建三個廣義結(jié)構(gòu)元對三種交通限速標(biāo)志進行擴展形態(tài)學(xué)濾波。根據(jù)構(gòu)建的廣義結(jié)構(gòu)元,對待識別圖像分別進行紅色圓形圖案、藍色圓形圖案、黑色圓形圖案檢測。(3)數(shù)字提取。根據(jù)檢測到的感興趣區(qū)域,在此區(qū)域中通過數(shù)字提取技術(shù),將交通限速標(biāo)志的關(guān)鍵數(shù)字提取出來并歸一化。(4)標(biāo)志識別。得到歸一化后的數(shù)字,根據(jù)三種交通限速標(biāo)志,分別采用對應(yīng)的基于16個特征分量、正確匹配數(shù)為10且對單個特征分量具有閾值限制的限速標(biāo)志識別模板進行識別。(5)實驗分析。采用C#語言開發(fā)實驗軟件,對本文算法理論進行驗證。通過實驗得到,本文設(shè)計的可識別殘損交通限速標(biāo)志的算法在對一般限速標(biāo)志和殘損限速標(biāo)志的識別正確率上分別為91.67%和89.83%,綜合識別正確率為90.80%。本文的主要創(chuàng)新點如下:(1)提出了利用廣義結(jié)構(gòu)元進行交通限速標(biāo)志檢測識別的方法。(2)提出了殘損交通限速標(biāo)志檢測識別算法。
[Abstract]:Intelligent Transportation system (its) focuses on transportation, service control and vehicle manufacturing, which aims at solving traffic congestion, enhancing vehicle safety and improving vehicle intelligence.Road traffic sign recognition (TSRs) is an important part of the research on service control in intelligent transportation system. The research of TSRs mainly needs electronic information, video processing, digital image processing, artificial intelligence and pattern recognition.At present, traffic speed limit sign detection and recognition methods generally have some problems, such as the recognition accuracy is not ideal, the recognition speed is not fast enough, and it is difficult to recognize the damaged traffic speed limit sign, and so on.Therefore, this paper chooses to study the detection and recognition method of traffic speed limit sign, adopts generalized structure element to detect the sign, designs a new template matching recognition algorithm, and proposes to solve the problem that the recognition rate is not high.The recognition speed is slow and it is difficult to identify the speed limit sign of damaged traffic.The main contents are as follows: 1) Image preprocessing.This paper compares the advantages and disadvantages of RGB and HSI space in detecting traffic speed limit signs, and selects HSI space to deal with traffic speed limit signs.In this space, histogram enhancement, Wiener filter, geometric transformation and other preprocessing.The idea of generalized structure element is introduced into the detection and identification of traffic speed limit sign, and three generalized structure elements are constructed to carry out extended morphological filtering for three traffic speed limit signs.According to the generalized structure element, the recognition image is extracted from red circular pattern, blue circular pattern and black circular pattern detection.According to the detected region of interest, the key numbers of traffic speed limit signs are extracted and normalized.The normalized number is obtained. According to the three traffic speed limit signs, the corresponding recognition template based on 16 feature components, with a correct matching number of 10 and a threshold limit for a single feature component, is used to identify the speed limit signs.The experimental software is developed with C # language, and the algorithm theory of this paper is verified.The experimental results show that the algorithm designed in this paper is 91.67% and 89.83% respectively for the recognition of the general speed limit sign and the damaged speed limit sign, and the comprehensive recognition accuracy rate is 90.80%.The main innovations of this paper are as follows: (1) A method of detecting and recognizing traffic speed limit signs using generalized structure elements is proposed.
【學(xué)位授予單位】：重慶交通大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2015
【分類號】：U495;TP391.41

【參考文獻】

相關(guān)期刊論文 前5條

1 張銘鈞;尚云超;楊杰;;基于灰度-梯度不變矩的水下目標(biāo)識別系統(tǒng)[J];哈爾濱工程大學(xué)學(xué)報;2009年06期

2 王建林;王志海;王學(xué)玲;;一種懶惰式?jīng)Q策樹和普通決策樹結(jié)合的分類模型——半懶惰式?jīng)Q策樹[J];計算機應(yīng)用與軟件;2008年12期

3 趙元慶;吳華;;多尺度特征和神經(jīng)網(wǎng)絡(luò)相融合的手寫體數(shù)字識別[J];計算機科學(xué);2013年08期

4 秦開懷;王海潁;鄭輯濤;;一種基于Hough變換的圓和矩形的快速檢測方法[J];中國圖象圖形學(xué)報;2010年01期

5 張燕昆;洪初陽;WANG Charles;;實時矩形交通限速標(biāo)志識別系統(tǒng)[J];智能系統(tǒng)學(xué)報;2010年06期

相關(guān)博士學(xué)位論文 前2條

1 張維澤;基于簡單局部特征學(xué)習(xí)的物體檢測方法[D];浙江大學(xué);2010年

2 藍章禮;橋梁健康監(jiān)測系統(tǒng)的數(shù)據(jù)獲�。豪碚摵头椒ㄑ芯縖D];重慶大學(xué);2008年

相關(guān)碩士學(xué)位論文 前1條

1 聶玖星;印刷體漢字識別系統(tǒng)的特征提取和匹配識別研究[D];大連理工大學(xué);2009年

，

本文編號：1687369