本文選題：數(shù)字圖像處理 + 人體動(dòng)作識(shí)別��； 參考：《電子科技大學(xué)》2015年博士論文

【摘要】：視頻人體動(dòng)作的分析與表示是計(jì)算機(jī)視覺(jué)領(lǐng)域的一個(gè)研究熱點(diǎn),其主要任務(wù)是從視頻中檢測(cè)、提取和表示人體運(yùn)動(dòng)信息,它涉及圖像處理、機(jī)器學(xué)習(xí)、應(yīng)用物理、數(shù)學(xué)等多個(gè)學(xué)科,具有重要的理論和實(shí)際應(yīng)用價(jià)值。由于人體運(yùn)動(dòng)的復(fù)雜性和多樣性,盡管經(jīng)歷了十幾年的研究,視頻人體動(dòng)作識(shí)別仍然難以應(yīng)用于實(shí)際環(huán)境。作為人體動(dòng)作識(shí)別的核心,動(dòng)作表示和識(shí)別仍然存在大量亟待解決的問(wèn)題。本文開(kāi)篇闡明了視頻人體動(dòng)作識(shí)別的研究背景、研究意義、主要任務(wù)以及典型模型,并從研究現(xiàn)狀及存在問(wèn)題兩個(gè)方面出發(fā),對(duì)運(yùn)動(dòng)檢測(cè)、特征提取及描述,編碼技術(shù)進(jìn)行了簡(jiǎn)單討論。在總結(jié)分析已有研究成果的基礎(chǔ)上,本文主要內(nèi)容包括四個(gè)方面:1)人體在時(shí)空中的運(yùn)動(dòng)會(huì)形成空間三維體,該三維體的形狀信息是重要的人體運(yùn)動(dòng)信息,這種形狀信息能夠被局部鄰域特征的位置關(guān)系所描述,為準(zhǔn)確描述這種關(guān)系,我們提出兩種局部鄰域特征構(gòu)造算法:基于正多面體的局部時(shí)空鄰域特征,和基于多尺度的時(shí)空方向鄰域特征。前者是利用正多面的多個(gè)空間軸作為特征位置的參考定位系統(tǒng),精確描述局部特征相對(duì)位置信息。后者是在局部鄰域的構(gòu)造中引入時(shí)空尺度參數(shù),使得鄰域特征具有方向選擇性。2)協(xié)方差特征是一種強(qiáng)有力的局部特征,本論文我們將局部人體運(yùn)動(dòng)信息表示為協(xié)方差特征,然后研究它在兩種情況下的動(dòng)作識(shí)別率:第一種情況,我們首先使用矩陣對(duì)數(shù)映射,將協(xié)方差從黎曼空間映射到Log-Euclidean空間,然后在Log-Euclidean空間進(jìn)行聚類(lèi)、編碼操作;第二種情況,為保持協(xié)方差特征在黎曼流形上的幾何結(jié)構(gòu)信息,我們直接對(duì)協(xié)方差矩陣在黎曼流形上進(jìn)行聚類(lèi)操作,生成黎曼矩陣字典,然后使用提出的局部黎曼流形編碼算法實(shí)現(xiàn)特征編碼。此外,我們還對(duì)不同矩陣距離度量下,協(xié)方差聚類(lèi)中的批量均值更新和順序均值更新做了深入研究。3)基于Grassmann隨機(jī)流形森林的人體動(dòng)作識(shí)別。傳統(tǒng)局部時(shí)空特征利用時(shí)空網(wǎng)格劃分局部時(shí)空體,然后分別計(jì)算每個(gè)網(wǎng)格的特征統(tǒng)計(jì)量,最后級(jí)聯(lián)所有網(wǎng)格的特征統(tǒng)計(jì)量,獲得局部特征描述子。這種網(wǎng)格劃分不僅破壞了幀與幀之間的時(shí)間關(guān)聯(lián)性,而且網(wǎng)格尺度沒(méi)有統(tǒng)一標(biāo)準(zhǔn),需要依靠經(jīng)驗(yàn)和實(shí)驗(yàn)確定。為解決該問(wèn)題,我們直接將每幀圖像拉成列向量,局部時(shí)空立方體被表示為列向量矩陣,為度量這些矩陣的相似度,我們使用Grassmann流形距離,然后利用Grassmann隨機(jī)流形樹(shù)描述Grassmann流形的數(shù)據(jù)概率分布信息和實(shí)現(xiàn)人體動(dòng)作分類(lèi)。4)特征編碼在動(dòng)作識(shí)別中占據(jù)重要地位,一直以來(lái)都是研究的熱點(diǎn)。我們通過(guò)對(duì)經(jīng)典局部約束線(xiàn)性編碼(Locality-constrained Linear Coding,LLC)算法的研究,提出一種LLC的加權(quán)版本,即WLLC編碼算法。LLC算法是近來(lái)提出的一種優(yōu)秀稀疏編碼,它的優(yōu)點(diǎn)包括編碼是稀疏的、編碼速度快、重構(gòu)誤差小,主要缺點(diǎn)是在其字典生成階段完全拋棄了數(shù)據(jù)聚類(lèi)中心附近樣本的概率分布信息,使得在編碼階段每個(gè)被選中的單詞對(duì)編碼的貢獻(xiàn)是一樣的。我們所提WLLC算法的基本思想是,由于每個(gè)單詞(聚類(lèi)中心)周?chē)?xùn)練樣本分布的差異,使得它們的可信度不同,在特征編碼中,高可信度的單詞應(yīng)該對(duì)編碼做出更大的貢獻(xiàn)。實(shí)驗(yàn)證明,通過(guò)引入WLLC編碼算法,動(dòng)作識(shí)別率被有效提高。此外,特征位置信息對(duì)于動(dòng)作識(shí)別具有重要意義,為此,我們提出一種混合特征,配合提出的多尺度空間位置編碼算法,達(dá)到準(zhǔn)確描述人體動(dòng)作在時(shí)空中的概率分布信息。論文最后對(duì)視頻人體動(dòng)作的分析與表示進(jìn)行了展望,并提出下一步工作的主要內(nèi)容。
[Abstract]:The analysis and representation of video human motion is a hot topic in the field of computer vision. Its main task is to detect, extract and express human motion information from video. It involves image processing, machine learning, applied physics, mathematics and other disciplines. It has important theory and practical application value. In spite of more than ten years of research, video human motion recognition is still difficult to apply to the actual environment. As the core of human action recognition, there are still a lot of problems to be solved in action representation and recognition. This paper expounds the research background, research significance, main tasks and typical models of video human action recognition. Based on the two aspects of the research status and existing problems, the motion detection, feature extraction and description and coding technology are discussed briefly. On the basis of summarizing and analyzing the existing research results, the main contents of this paper include four aspects: 1) the movement of the human body in time and space will form a three-dimensional body, and the shape information of the three-dimensional body is heavy. For human motion information, this shape information can be described by the location relationship of local neighborhood features. In order to accurately describe the relationship, we propose two local neighborhood feature construction algorithms, based on the local spatiotemporal neighborhood features of the positive polyhedron, and the multi-scale based spatio-temporal neighborhood characteristics. As a reference location system, the spatial axis accurately describes the relative position information of local features. The latter is the introduction of time and space parameters in the construction of local neighborhood, which makes the neighborhood features.2) covariance feature is a powerful local feature. In this paper, we represent the local human motion information as a covariance. Characteristic of variance, and then study the motion recognition rate in two cases: first, we first use matrix logarithmic mapping to map covariance from Riemann space to Log-Euclidean space, then cluster, code operation and second conditions in Log-Euclidean space to keep the geometric knot of covariance features on the Riemann manifold. In order to construct the information, we directly cluster the covariance matrix on the Riemann manifold, generate the Riemann matrix dictionary, and then use the proposed local Riemann manifold coding algorithm to implement the feature coding. In addition, we also make a thorough study of the mean renewal of the batch and the sequential mean renewal in the covariance clustering under the different matrix distance metrics.3 ) the human movement recognition based on the Grassmann random manifold forest. The traditional local spatiotemporal features are divided into the local space-time bodies using the space-time grid, then the characteristic statistics of each grid are calculated respectively. Finally, the feature statistics of all the grids are concatenated, and the local feature descriptors are obtained. This mesh division not only destroys the time between frames and frames. In order to solve the problem, we directly pull each frame into a column vector, and the local space-time cube is expressed as a column vector matrix to measure the similarity of these matrices. We use the Grassmann manifold distance and then use the Grassmann random manifold tree to describe the problem. The data probability distribution information of Grassmann manifolds and the implementation of human action classification.4) feature encoding play an important role in the action recognition. It has always been the hot spot of research. By the study of the classical local constrained linear coding (Locality-constrained Linear Coding, LLC), a weighted version of LLC, that is, WLLC, is proposed. The coding algorithm.LLC algorithm is an excellent sparse coding which has been proposed recently. Its advantages include the sparse coding, fast coding speed and small reconstruction error. The main disadvantage is that the probability distribution information of the samples near the data cluster center is completely abandoned in the phase of the dictionary generation, so that each selected word is coded at the coding stage. The contribution is the same. The basic idea of the WLLC algorithm we propose is that because of the differences in the distribution of training samples around each word (cluster center), their credibility is different. In feature coding, high reliability words should make a greater contribution to the coding. The experimental evidence shows that the action recognition rate is introduced by introducing the WLLC coding algorithm. In addition, the feature location information is of great significance to the action recognition. For this reason, we propose a hybrid feature, combined with the proposed multi-scale spatial location coding algorithm, to accurately describe the probability distribution information of human movements in time and space. The main content of the next step of the work.

【學(xué)位授予單位】：電子科技大學(xué)
【學(xué)位級(jí)別】：博士
【學(xué)位授予年份】：2015
【分類(lèi)號(hào)】：TP391.41

【共引文獻(xiàn)】

相關(guān)期刊論文 前10條

1 陳文;基于決策樹(shù)的入侵檢測(cè)的實(shí)現(xiàn)[J];安徽技術(shù)師范學(xué)院學(xué)報(bào);2005年05期

2 彭莉芬;陳俊生;胡學(xué)鋼;;基于粗糙集決策樹(shù)算法的研究[J];安慶師范學(xué)院學(xué)報(bào)(自然科學(xué)版);2012年01期

3 趙玉鵬;;論機(jī)器學(xué)習(xí)[J];安陽(yáng)工學(xué)院學(xué)報(bào);2011年04期

4 孫雪;李昆侖;胡夕坤;趙瑞;;基于半監(jiān)督K-means的K值全局尋優(yōu)算法[J];北京交通大學(xué)學(xué)報(bào);2009年06期

5 趙勇;劉凱;;數(shù)字挖掘方法在遙感分類(lèi)中的應(yīng)用研究[J];北京測(cè)繪;2009年03期

6 沈奕,滑峰,劉椿年;基于GDT的對(duì)FOIL系統(tǒng)的改進(jìn)[J];北京工業(yè)大學(xué)學(xué)報(bào);2005年02期

7 朱青;劉宇輝;;一種面向領(lǐng)域的組件質(zhì)量度量算法[J];北京工業(yè)大學(xué)學(xué)報(bào);2007年01期

8 陳陽(yáng)舟;黃旭;代桂平;;基于新的狀態(tài)劃分的多機(jī)器人圍捕策略[J];北京工業(yè)大學(xué)學(xué)報(bào);2010年08期

9 張瑞華;周延泉;王樅;李蕾;;移動(dòng)終端離線(xiàn)瀏覽系統(tǒng)的新聞推薦服務(wù)研究[J];北京郵電大學(xué)學(xué)報(bào);2006年06期

10 楊種學(xué);;基于回歸技術(shù)商品銷(xiāo)售趨勢(shì)預(yù)測(cè)模型的實(shí)現(xiàn)[J];保山師專(zhuān)學(xué)報(bào);2009年05期

相關(guān)博士學(xué)位論文 前10條

1 全惠敏;電能質(zhì)量相關(guān)信號(hào)的S變換檢測(cè)算法及應(yīng)用研究[D];湖南大學(xué);2010年

2 高山;蛋白質(zhì)點(diǎn)突變效果預(yù)測(cè)與突變數(shù)據(jù)庫(kù)研究[D];南開(kāi)大學(xué);2010年

3 曹葵康;支持向量機(jī)加速方法及應(yīng)用研究[D];浙江大學(xué);2010年

4 林龍信;仿生水下機(jī)器人的增強(qiáng)學(xué)習(xí)控制方法研究[D];國(guó)防科學(xué)技術(shù)大學(xué);2010年

5 杜偉;機(jī)器學(xué)習(xí)及數(shù)據(jù)挖掘在生物信息學(xué)中的應(yīng)用研究[D];吉林大學(xué);2011年

6 聶黎;基于基因表達(dá)式編程的車(chē)間動(dòng)態(tài)調(diào)度方法研究[D];華中科技大學(xué);2011年

7 楊抒;基于WEB的林產(chǎn)品信息資源整合方法研究[D];北京林業(yè)大學(xué);2011年

8 黃靜華;支持向量機(jī)算法研究及在氣象數(shù)據(jù)挖掘中的應(yīng)用[D];中國(guó)礦業(yè)大學(xué)（北京）;2011年

9 陳俊;笑臉表情分類(lèi)識(shí)別的研究[D];華南理工大學(xué);2011年

10 劉宏兵;多目標(biāo)粒度支持向量機(jī)及其應(yīng)用研究[D];武漢理工大學(xué);2011年

，

本文編號(hào)：1787483