本文選題：智能監(jiān)控系統(tǒng)　切入點(diǎn)：目標(biāo)跟蹤　出處：《蘭州理工大學(xué)》2014年碩士論文　論文類型：學(xué)位論文

【摘要】：以計(jì)算機(jī)視覺(jué)技術(shù)為基礎(chǔ)的智能視頻監(jiān)控系統(tǒng),目前已經(jīng)廣泛應(yīng)用于人們生產(chǎn)生活的各個(gè)方面。智能監(jiān)控是集成了智能行為識(shí)別算法,能夠?qū)Ξ?huà)面場(chǎng)景中的人或車輛的行為進(jìn)行識(shí)別、判斷,并在適當(dāng)?shù)臈l件下,產(chǎn)生報(bào)警提示用戶；運(yùn)動(dòng)目標(biāo)檢測(cè)與跟蹤是智能視頻監(jiān)控系統(tǒng)中的關(guān)鍵技術(shù)之一。本文主要研究了智能視頻監(jiān)控系統(tǒng)中目標(biāo)跟蹤技術(shù)的實(shí)現(xiàn)方法,其主要研究工作如下： 1.首先對(duì)傳統(tǒng)智能視頻監(jiān)控系統(tǒng)中常用的3種目標(biāo)跟蹤算法(卡爾曼濾波算法、均值漂移算法、自適應(yīng)均值漂移算法)進(jìn)行了研究；然后,通過(guò)分別在不同遮擋情況下(無(wú)遮擋、半遮擋、全遮擋等),對(duì)3種目標(biāo)跟蹤算法進(jìn)行了仿真實(shí)驗(yàn)；并通過(guò)對(duì)比分析,歸納出了3種跟蹤算法的優(yōu)點(diǎn)以及存在的問(wèn)題。 2.針對(duì)自適應(yīng)均值漂移算法在目標(biāo)發(fā)生遮擋時(shí)魯棒性差的問(wèn)題,提出了一種基于自適應(yīng)均值漂移的改進(jìn)的目標(biāo)跟蹤算法�；诳柭鼮V波算法有較好的預(yù)測(cè)特性以及自適應(yīng)均值漂移算法具有較高的實(shí)時(shí)性的基礎(chǔ)上,本文將卡爾曼濾波算法應(yīng)用于自適應(yīng)均值漂移算法中,來(lái)解決自適應(yīng)均值算法當(dāng)目標(biāo)發(fā)生遮擋時(shí)跟蹤失敗的問(wèn)題；在目標(biāo)跟蹤過(guò)程中通過(guò)巴氏系數(shù)來(lái)判斷目標(biāo)是否被遮擋,當(dāng)目標(biāo)出現(xiàn)遮擋時(shí)運(yùn)用卡爾曼濾波進(jìn)行預(yù)測(cè),然后把預(yù)測(cè)的結(jié)果作為自適應(yīng)均值漂移算法的下一次輸入,在目標(biāo)出現(xiàn)對(duì)其進(jìn)行重新快速的捕獲。仿真實(shí)驗(yàn)結(jié)果證明,本文所提出的改進(jìn)算法具有較好的實(shí)時(shí)性,且在目標(biāo)遮擋情況下具有較強(qiáng)的魯棒性。 3.在Microsoft Visual Studio2008集成開(kāi)發(fā)環(huán)境下,采用QT應(yīng)用程序框架及OpenCV計(jì)算機(jī)視覺(jué)庫(kù)代碼,實(shí)現(xiàn)了基于運(yùn)動(dòng)目標(biāo)檢測(cè)與跟蹤的視頻監(jiān)控系統(tǒng)。該系統(tǒng)對(duì)于USB攝像頭或AVI視頻文件輸入的視頻,能實(shí)時(shí)檢測(cè)出場(chǎng)景中的運(yùn)動(dòng)物體并進(jìn)行跟蹤。
[Abstract]:Intelligent video monitoring system based on computer vision technology, has been widely used in all aspects of people's life and production. The intelligent monitoring system integrated intelligent behavior recognition algorithms to picture the scene in person or vehicle behavior recognition, judgment, and under appropriate conditions, an alarm prompts the user; moving target detection and tracking is one of the key technologies in intelligent video surveillance system. This paper mainly studies the method to realize the target tracking technology in intelligent video surveillance system, the main research work is as follows:
1. the first of the 3 objectives of the traditional intelligent video surveillance system tracking algorithm (Calman filtering algorithm, mean shift algorithm, adaptive mean shift algorithm) is studied; then, through respectively in different occlusion condition (no occlusion, half occlusion, occlusion, etc.) of 3 kinds of target tracking algorithm experiment; and through comparative analysis, summed up the advantages of the 3 kinds of tracking algorithm and existing problems.
2. adaptive mean shift algorithm in target robust occlusion problems, put forward a kind of improved adaptive mean shift tracking algorithm based on target. Calman filtering algorithm based on prediction and better performance of adaptive mean shift algorithm has high real-time on the basis of the Calman filter algorithm applied to adaptive mean shift algorithm, to solve the adaptive k-means algorithm when the target is occluded by tracking failure problem; Bhattacharyya coefficient to judge whether the target is occluded in the target tracking process, when the target when there is occlusion is predicted using the Calman filter, and then the forecast result as the adaptive mean shift algorithm for the next input in the target there re quick to capture it. The simulation results show that the algorithm proposed in this paper has better It is real-time and has strong robustness in the case of target occlusion.
3. in the Microsoft Visual Studio2008 integrated development environment, using QT application framework and OpenCV computer vision library code, the realization of the video monitoring system for moving target detection and tracking based on the input system for the USB camera or AVI video video files, can be used to detect moving objects in the scene and track.

【學(xué)位授予單位】：蘭州理工大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2014
【分類號(hào)】：TN948.6

【參考文獻(xiàn)】

相關(guān)博士學(xué)位論文 前1條

1 王愛(ài)平;視頻目標(biāo)跟蹤技術(shù)研究[D];國(guó)防科學(xué)技術(shù)大學(xué);2011年

，

本文編號(hào)：1561234