發(fā)布時間：2018-06-08 05:07

  本文選題：高速公路 + 偏色檢測　； 參考：《重慶大學(xué)》2015年碩士論文

【摘要】：視頻監(jiān)控系統(tǒng)是高速公路管理系統(tǒng)中的重要組成部分,監(jiān)控畫面清晰正常是監(jiān)控效果的重要指標(biāo)。然而,實際的監(jiān)控視頻往往存在信號缺失、偏色、模糊、攝像頭干擾等異常,影響視頻監(jiān)控系統(tǒng)的效果,且不利于后續(xù)高速公路異常事件的檢測。現(xiàn)有的視頻異常檢測方法受復(fù)雜環(huán)境影響、實時性差,難以滿足復(fù)雜的高速公路監(jiān)控場景和實時性檢測要求。為此,利用數(shù)字圖像處理技術(shù),實現(xiàn)高速公路監(jiān)控視頻異常的自動檢測具有重要的學(xué)術(shù)意義和應(yīng)用價值。針對上述問題,本文重點研究了視頻圖像偏色和攝像頭干擾的檢測技術(shù),以及對不符合高速公路監(jiān)控錄像質(zhì)量標(biāo)準(zhǔn)的視頻自動篩選和判別技術(shù)。針對高速公路視頻監(jiān)控系統(tǒng)中經(jīng)常出現(xiàn)的視頻信號缺失和監(jiān)控畫面凍結(jié)的信號故障問題,本文分別給出了基于圖像像素灰度值的方差、幀差法來實現(xiàn)這兩種信號故障問題的檢測。重點針對圖像偏色問題,提出了一種偏色因子的計算方法,并給出了基于RGB三維顏色空間直角坐標(biāo)系的偏色檢測方法。通過在RGB顏色空間下,向量化處理每個顏色分量并計算其偏色因子實現(xiàn)對視頻監(jiān)控系統(tǒng)中偏色異常事件的檢測。實驗結(jié)果表明,該檢測方法能夠有效的檢測出偏色的監(jiān)控視頻。針對高速公路攝像頭經(jīng)常發(fā)生平移、偏轉(zhuǎn)、遮擋等干擾問題,本文提出了基于動態(tài)Harris角點模板匹配的高速公路攝像頭干擾檢測方法。利用角點在描述圖像位置形狀等方面的優(yōu)勢,基于掩膜提取技術(shù),采用三張構(gòu)造的灰度圖對特定檢測區(qū)域內(nèi)的角點進(jìn)行提取,然后根據(jù)角點數(shù)量和位置信息提出動態(tài)角點模板匹配算法計算匹配因子,通過匹配因子實現(xiàn)對攝像頭的干擾檢測。與Evan Ribnick檢測方法相比,所提檢測算法不僅耗時少,且具有較強的適應(yīng)性和抗干擾能力。針對高速公路監(jiān)控錄像中常有的偏暗、偏白、模糊等不合格視頻,本文提出了基于角點鄰域像素標(biāo)準(zhǔn)差的視頻圖像清晰度評價方法來實現(xiàn)對高速公路視頻圖像質(zhì)量的異常檢測。該方法首先通過計算平均能量強度來排除偏暗或偏白的視頻,再對視頻圖像結(jié)構(gòu)特征進(jìn)行分析,然后基于提出的視頻圖像清晰度評價方法對視頻圖像的清晰度進(jìn)行評價,排除模糊的監(jiān)控視頻。實測高速公路視頻驗證了該方法的一致性和穩(wěn)定性。最后,利用上述視頻異常檢測算法完成了高速公路監(jiān)控視頻異常檢測系統(tǒng)的設(shè)計和實現(xiàn),并編寫了測試軟件,對算法模塊性能進(jìn)行了測試,并對高速公路隧道、關(guān)鍵路段、收費廣場場景下的監(jiān)控視頻進(jìn)行了檢測。實驗結(jié)果表明,該系統(tǒng)能夠較為正確地檢測出高速公路監(jiān)控異常視頻,并能滿足檢測的實時性要求。
[Abstract]:Video surveillance system is an important part of highway management system. However, the actual surveillance video often has abnormal signal, color deviation, blur, camera interference and so on, which affects the effect of video surveillance system, and is not conducive to the subsequent detection of highway abnormal events. The existing video anomaly detection methods are affected by complex environment and have poor real-time performance, so it is difficult to meet the requirements of complex freeway monitoring scene and real-time detection. Therefore, it is of great academic significance and application value to use digital image processing technology to realize automatic detection of highway surveillance video anomalies. Aiming at the above problems, this paper focuses on the detection technology of video image color deviation and camera interference, as well as the automatic video screening and discrimination technology which does not meet the quality standard of highway surveillance video. In order to solve the problems of missing video signal and frozen image signal in freeway video surveillance system, the variance based on pixel gray value of image is given in this paper. Frame difference method is used to detect the two kinds of signal faults. Aiming at the problem of image color deviation, this paper puts forward a method to calculate the color deviation factor, and gives a method of color deviation detection based on RGB 3D color space right angle coordinate system. In the RGB color space, each color component is processed by vectorization and its coloration factor is calculated to detect abnormal events in video surveillance system. Experimental results show that the detection method can effectively detect the color deviation of surveillance video. Aiming at the interference problems of highway camera, such as translation, deflection, occlusion and so on, this paper presents an interference detection method for freeway camera based on dynamic Harris corner template matching. Taking advantage of corner points in describing image position and shape, based on mask extraction technology, three gray scale images are used to extract corner points in a specific detection region. Then according to the number of corner points and location information, a dynamic corner template matching algorithm is proposed to calculate the matching factor, and the interference detection of camera is realized by matching factor. Compared with Evan Ribnick detection method, the proposed detection algorithm not only takes less time, but also has strong adaptability and anti-jamming ability. Aiming at the substandard video such as dark, white and fuzzy in highway surveillance video, this paper proposes a method of video image definition evaluation based on corner neighborhood pixel standard deviation to realize the abnormal detection of highway video image quality. Firstly, the average energy intensity is calculated to eliminate the dark or white video, then the structure features of the video image are analyzed, and then the definition of the video image is evaluated based on the proposed method. Remove fuzzy surveillance video. The consistency and stability of the method are verified by the actual highway video. Finally, the design and implementation of the video anomaly detection system for expressway surveillance is completed by using the above video anomaly detection algorithm, and the test software is written, the performance of the algorithm module is tested, and the highway tunnel and key sections are tested. The surveillance video of the toll square scene was detected. The experimental results show that the system can detect the abnormal video of expressway monitoring correctly and can meet the real-time requirements of the detection.
【學(xué)位授予單位】：重慶大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2015
【分類號】：TN948.6;U495

【相似文獻(xiàn)】

相關(guān)期刊論文 前10條

1 謝俊宏;;淺談監(jiān)控視頻干擾問題的解決[J];海峽科學(xué);2008年08期

2 歐溫暖;田緒紅;朱同林;;基于教室監(jiān)控視頻的課堂行為計數(shù)分析[J];圖學(xué)學(xué)報;2012年02期

3 李勃;陳啟美;;基于監(jiān)控視頻的運動車輛行為分析算法[J];儀器儀表學(xué)報;2006年S3期

4 梁浩哲;徐樹奎;李國輝;張軍;;面向監(jiān)控視頻的行為模式挖掘[J];山東大學(xué)學(xué)報(工學(xué)版);2011年06期

5 孫晉明;;遠(yuǎn)程監(jiān)控視頻無線傳輸方式的比選設(shè)計[J];電氣時代;2011年05期

6 姜振鳳;遲慶云;;基于DBAN的監(jiān)控視頻數(shù)據(jù)挖掘[J];河北工程大學(xué)學(xué)報(自然科學(xué)版);2013年02期

7 祝林峰;方思平;張興明;張平;馬旭;;監(jiān)控視頻中的車體特征檢驗在交通肇事逃逸案件中的應(yīng)用[J];警察技術(shù);2013年06期

8 胡平;;霧霾天氣下高速公路監(jiān)控視頻的一種增強算法[J];公路交通科技;2014年01期

9 閆帥輝;柏正堯;尹立國;楊仁青;高v，

本文編號：1994624