本文關(guān)鍵詞：智能視頻監(jiān)控系統(tǒng)的運動目標檢測與跟蹤算法研究　出處：《寧夏大學(xué)》2014年碩士論文　論文類型：學(xué)位論文

  更多相關(guān)文章： 智能視頻監(jiān)控 運動目標檢測 陰影檢測與去除 運動目標跟蹤 顏色空間轉(zhuǎn)換 卡爾曼濾波

【摘要】：安防理念的不斷提高使智能視頻監(jiān)控在人類社會生活中扮演著越來越重要的角色。智能視頻監(jiān)控的核心任務(wù)是對視頻流中的圖像序列進行有效分析,實現(xiàn)場景中運動目標的檢測和跟蹤,以及完成后續(xù)的目標識別、行為理解等任務(wù),其中大部分有意義的信息體現(xiàn)在運動中,故運動目標的檢測與跟蹤是目前智能視頻監(jiān)控系統(tǒng)中最重要的研究內(nèi)容之一。本文分三個方面圍繞運動目標檢測和跟蹤算法開展深入研究,最后闡述了智能視頻監(jiān)控系統(tǒng)的工作原理、技術(shù)架構(gòu)及其在居家安防中的應(yīng)用情況、優(yōu)越性、關(guān)鍵技術(shù)及難點和未來發(fā)展趨勢。 (一)對幾種常用的運動目標檢測方法進行分析,選用混合高斯模型進行背景建模,并對檢測出的前景目標選取了合適的形態(tài)學(xué)操作進行去噪。 (二)針對當(dāng)前運動陰影檢測中采用的紋理信息過于粗糙、閾值選取需要人工干涉等問題,通過對NCC(歸一化互相關(guān))紋理算法進行改進,并結(jié)合亮度和歸一化顏色特性,提出一種自適應(yīng)的運動陰影檢測方法,以混合高斯模型得到的前景像素為基礎(chǔ),通過陰影在亮度和歸一化顏色的特性篩選出候選的陰影區(qū)域,結(jié)合改進的紋理算法進一步縮小陰影區(qū)域范圍,最后利用空間后處理得到真實陰影。實驗結(jié)果表明,改進后的算法在有效降低噪聲干擾的情況下能夠較好的區(qū)分局部紋理不明顯的運動目標和陰影。 (三)針對傳統(tǒng)Meanshift算法在運動目標被嚴重遮擋情況下出現(xiàn)跟蹤丟失問題,提出了復(fù)雜環(huán)境中融合軌跡校正的新型Meanshift目標跟蹤算法。將顏色空間由傳統(tǒng)的RGB空間轉(zhuǎn)換到區(qū)分度更好的HSV空間,提出了新的融合規(guī)則：目標無遮擋和走出遮擋時,Meanshift算法進行跟蹤；目標進入遮擋和被嚴重或完全遮擋時,Klaman濾波估計運動軌跡。實驗結(jié)果表明：新算法有效解決了目標處于遮擋下的跟蹤丟失問題。 (四)對智能視頻監(jiān)控系統(tǒng)的工作原理、體系架構(gòu)進行分析研究,簡要闡述了智能視頻監(jiān)控系統(tǒng)在人們?nèi)粘Ｉ钪械闹匾院蛢?yōu)越性。重點對家庭智能視頻監(jiān)控系統(tǒng)做了深入的研究,分析其現(xiàn)實應(yīng)用情況、技術(shù)優(yōu)勢、關(guān)鍵技術(shù)及難點和未來的發(fā)展趨勢。 本文的研究內(nèi)容較好地提高了運動目標的陰影檢測能力,解決了遮擋時的跟蹤丟失問題,對于視頻監(jiān)控系統(tǒng)中的目標陰影檢測和跟蹤有一定的應(yīng)用價值和指導(dǎo)意義。
[Abstract]:With the constant improvement of security concept, intelligent video surveillance plays an increasingly important role in human social life. The core task of intelligent video surveillance is to effectively analyze the image sequence in video stream. To achieve the detection and tracking of moving targets in the scene, as well as to complete the tasks of target recognition, behavior understanding, and so on, most of the meaningful information is reflected in the motion. So the detection and tracking of moving targets is one of the most important research contents in the intelligent video surveillance system. Finally, the working principle, technical framework and application in home security of intelligent video surveillance system are described. The advantages, key technologies, difficulties and future development trends are also discussed. (1) based on the analysis of several commonly used moving target detection methods, the mixed Gao Si model is used to model the background, and the suitable morphological operation is selected for denoising the detected foreground target. (2) aiming at the problem that the texture information used in the current motion shadow detection is too rough and the threshold selection needs manual interference, the NCC (normalized cross-correlation) texture algorithm is improved. Combined with brightness and normalized color characteristics, an adaptive moving shadow detection method is proposed, which is based on foreground pixels obtained by mixed Gao Si model. The candidate shadow region is selected by the feature of shadow brightness and normalized color, and the shadow area is further reduced by the improved texture algorithm. The experimental results show that the improved algorithm can effectively reduce the noise interference and distinguish the moving object from the shadow which the local texture is not obvious. (3) aiming at the problem of tracking loss in the case that the moving object is seriously occluded, the traditional Meanshift algorithm appears. A new Meanshift target tracking algorithm based on fusion trajectory correction in complex environments is proposed. The color space is transformed from the traditional RGB space to a better discriminant HSV space. A new fusion rule is proposed: target without occlusion and out of occlusion with mean shift algorithm for tracking; The Klaman filter is used to estimate the motion trajectory when the target is occluded and severely or completely occluded. The experimental results show that the new algorithm can effectively solve the tracking loss problem of the target under occlusion. (4) the working principle and architecture of intelligent video surveillance system are analyzed and studied. This paper briefly expounds the importance and superiority of intelligent video surveillance system in people's daily life. It focuses on the in-depth study of home intelligent video surveillance system and analyzes its practical application and technical advantages. Key technologies and difficulties and future trends. The research content of this paper improves the shadow detection ability of moving targets and solves the problem of tracking loss in occlusion. It has certain application value and guiding significance for target shadow detection and tracking in video surveillance system.
【學(xué)位授予單位】：寧夏大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2014
【分類號】：TP391.41;TN948.6

【參考文獻】

相關(guān)期刊論文 前10條

1 譚熊;余旭初;劉景正;黃偉杰;;基于無人機視頻的運動目標快速跟蹤[J];測繪通報;2011年09期

2 程磊;;運動目標檢測跟蹤方法研究[J];電腦知識與技術(shù);2010年07期

3 陳延濤;熊運余;唐飛;楊璇;李鵬宏;;基于區(qū)域統(tǒng)計及YUV色度特性的陰影消除[J];光電子.激光;2009年09期

4 高俊祥;田彥;劉勇;;智能交通系統(tǒng)中的橢球法運動陰影檢測[J];光電子.激光;2009年10期

5 許曉航;肖剛;云霄;謝金華;;復(fù)雜背景及遮擋條件下的運動目標跟蹤[J];光電工程;2013年01期

6 趙謙;朱華偉;曾召華;侯媛彬;;基于YUV顏色空間特征的目標跟蹤融合算法[J];電視技術(shù);2013年09期

7 王樹文,閆成新,張?zhí)煨?趙廣州;數(shù)學(xué)形態(tài)學(xué)在圖像處理中的應(yīng)用[J];計算機工程與應(yīng)用;2004年32期

8 王展青;凡友福;張桂林;;跟蹤遮擋目標的一種魯棒算法[J];計算機工程與應(yīng)用;2007年27期

9 任建強;;一種自適應(yīng)的運動目標陰影消除新算法[J];計算機工程與應(yīng)用;2010年36期

10 林慶;陳遠祥;王士同;詹永照;;窗口尺寸變化目標的遮擋跟蹤[J];計算機科學(xué);2010年08期

相關(guān)博士學(xué)位論文 前2條

1 袁國武;智能視頻監(jiān)控中的運動目標檢測和跟蹤算法研究[D];云南大學(xué);2012年

2 周維;視頻監(jiān)控中運動目標發(fā)現(xiàn)與跟蹤算法研究[D];中國科學(xué)技術(shù)大學(xué);2012年

，

本文編號：1440600