發(fā)布時間：2018-04-25 15:08

  本文選題：視頻監(jiān)控 + 寬景拼接��； 參考：《西南交通大學(xué)》2014年碩士論文

【摘要】：伴隨信息化技術(shù)的發(fā)展和普及,視頻監(jiān)控系統(tǒng)己在各個領(lǐng)域得到了廣泛的重視和運用。傳統(tǒng)的視頻監(jiān)控大都以單個攝像頭為節(jié)點與相連的電腦組成,這種監(jiān)控系統(tǒng)視野相對來說比較局限,多數(shù)情況下存在著監(jiān)控死角,監(jiān)控效果往往不理想。而傳統(tǒng)的寬景視頻監(jiān)控多數(shù)是基于個人電腦或?qū)Ｓ玫膱D像處理板的,其體積大、功耗高、隱蔽性差。 寬景視頻監(jiān)控系統(tǒng),將傳統(tǒng)的多個攝像頭監(jiān)控內(nèi)容進(jìn)行整合,采用圖像拼接算法以及數(shù)字圖像處理技術(shù)去除圖像中的重疊區(qū)域,輸出一種視野更加開闊的寬景視頻數(shù)據(jù)流,從而大大增加了傳統(tǒng)的監(jiān)控范圍,減小了監(jiān)控死角的出現(xiàn)。因此,越來越受到人們的青睞。于此同時,基于嵌入式平臺的監(jiān)控系統(tǒng)具有成本低、體積小、功耗低、隱蔽性好的特性,克服了傳統(tǒng)視頻監(jiān)控的多個不足。 本文以Linux為核心操作系統(tǒng),選用以德州儀器公司(Texas Instruments, TI)最新推出的TMS320DM8168為核心的高性能嵌入式設(shè)備SEED-DVS8168為開發(fā)平臺。利用SEED-DVS8168平臺拓展的多路數(shù)據(jù)接口,同時采集兩路場景視頻內(nèi)容進(jìn)行視頻拼接處理,獲得比單個攝像頭視野增加一倍的寬景視頻輸出顯示。本文充分利用DM8168處理器內(nèi)部集成的ARM處理器、數(shù)字信號處理(Digital Singnal Processor, DSP)以及圖像處理協(xié)處理器的協(xié)同合作,完成了視頻的預(yù)處理、視頻拼接、視頻壓縮并組成了新的視頻流以輸出顯示等過程。與傳統(tǒng)的單核視頻或ARM+DSP異構(gòu)多核的視頻監(jiān)控不同的是,本文結(jié)合硬件平臺的結(jié)構(gòu)特點發(fā)揮各個子模塊的特點,將系統(tǒng)劃分為多個子系統(tǒng)進(jìn)行模塊化管理,把視頻從采集到顯示的各個過程分布到各個子系統(tǒng),減小了傳統(tǒng)監(jiān)控單機處理的運算負(fù)載,大大提升了本系統(tǒng)的整體效率。本文重點對多系統(tǒng)之間的協(xié)同工作、消息通信以及數(shù)據(jù)共享的機制進(jìn)行研究設(shè)計,同時針對多路圖像采集的同步問題進(jìn)行了處理。同傳統(tǒng)視頻監(jiān)控相比,這種設(shè)計的寬景視頻監(jiān)控系統(tǒng)在相同的畫面中提供更加豐富的視野空間和視覺內(nèi)容,擁有者更好的用戶體驗。這種基于寬景的視頻監(jiān)控系統(tǒng)可應(yīng)用在汽車安全駕駛、大型會場的監(jiān)控、十字路口監(jiān)控、樓盤介紹等諸多領(lǐng)域,這類對過程監(jiān)控和大場景監(jiān)控要求較高的監(jiān)控場所特別適合使用寬景視頻監(jiān)控,并且具有低成本、低功耗、穩(wěn)定性強、隱蔽性高等應(yīng)用特點,因此具有很好的研究價值和實用價值。
[Abstract]:With the development and popularization of information technology, video surveillance system has been widely paid attention to and applied in various fields. Traditional video surveillance mostly consists of a single camera node and a connected computer. The visual field of this kind of monitoring system is relatively limited. In most cases, there is a monitoring dead angle, and the monitoring effect is often not ideal. Traditional wide-view video surveillance is based on personal computer or special image processing board, which is large in size, high in power consumption and poor in concealment. The wide-view video surveillance system integrates the traditional multi-camera monitoring content, uses the image mosaic algorithm and the digital image processing technology to remove the overlapping area in the image, and outputs a broad-view video data stream with a wider field of vision. Thus, the traditional monitoring range is greatly increased and the dead angle of monitoring is reduced. Therefore, more and more people favor. At the same time, the monitoring system based on embedded platform has the characteristics of low cost, small volume, low power consumption and good concealment, which overcomes many shortcomings of traditional video surveillance. In this paper, Linux is used as the core operating system, and the high performance embedded device SEED-DVS8168, which is the core of Texas Instruments-Texas Instruments (tie), is chosen as the development platform. Using the multi-channel data interface extended by the SEED-DVS8168 platform, two scene video contents are collected simultaneously for video mosaic processing, and the wide-view video output display is obtained, which is twice as large as the single camera field of view. In this paper, we make full use of the cooperation of ARM processor, Digital Singnal processor (DSP) and image processing coprocessor (DSP) to complete video preprocessing and video splicing. Video compression and composition of a new video stream to output display and other processes. Different from the traditional single-core video or ARM DSP heterogeneous multi-core video surveillance, this paper combines the structure of the hardware platform to give play to the characteristics of each sub-module, the system is divided into multiple subsystems for modular management. The video process from acquisition to display is distributed to each subsystem, which reduces the computational load of the traditional monitoring and control system, and greatly improves the overall efficiency of the system. In this paper, we focus on the research and design of cooperative work, message communication and data sharing among multi-systems. At the same time, we deal with the synchronization of multi-channel image acquisition. Compared with traditional video surveillance, the designed wide-view video surveillance system provides richer visual space and visual content in the same picture, and the owner has a better user experience. This kind of video surveillance system based on wide view can be used in many fields, such as safe driving of cars, monitoring of large meeting places, intersections monitoring, introduction of buildings, etc. This kind of monitoring places with high requirements for process monitoring and large scene monitoring are especially suitable for wide-view video surveillance, and have the characteristics of low cost, low power consumption, strong stability, high concealment and so on. Therefore, it has good research value and practical value.
【學(xué)位授予單位】：西南交通大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2014
【分類號】：TN948.6;TP391.41

【參考文獻(xiàn)】

相關(guān)期刊論文 前10條

1 王啟春;郭廣禮;查劍鋒;劉神;;基于圖像灰度點特征提取算子的比較研究及改進(jìn)[J];大地測量與地球動力學(xué);2012年02期

2 王建平;季學(xué)鋒;穆道明;;基于DSP的多路音/視頻采集處理系統(tǒng)設(shè)計[J];國外電子元器件;2006年06期

3 楊宇博;程承旗;;基于分塊Harris特征的航拍視頻拼接方法[J];北京大學(xué)學(xué)報(自然科學(xué)版);2013年04期

4 張華俊;;實時視頻拼接系統(tǒng)[J];大眾科技;2012年04期

5 賈平;徐寧;張葉;;基于局部特征提取的目標(biāo)自動識別[J];光學(xué)精密工程;2013年07期

6 趙付軒;楊斌;;基于異構(gòu)雙核平臺的多媒體算法研究[J];單片機與嵌入式系統(tǒng)應(yīng)用;2012年11期

7 曹世翔;江潔;張廣軍;袁艷;;邊緣特征點的多分辨率圖像拼接[J];計算機研究與發(fā)展;2011年09期

8 王小強;陳臨強;梁旭;;實時全自動視頻拼接方法[J];計算機工程;2011年05期

9 李軍;吳潔明;;一種改進(jìn)圖像拼接算法的仿真研究[J];計算機仿真;2012年02期

10 徐賢鋒;檀結(jié)慶;;一種改進(jìn)的多尺度Harris特征點檢測方法[J];計算機工程;2012年17期

相關(guān)博士學(xué)位論文 前1條

1 邵向鑫;數(shù)字圖像拼接核心算法研究[D];吉林大學(xué);2010年

，

本文編號：1801844