本文選題：人體姿態(tài)識別　切入點：深度圖像　出處：《成都理工大學》2017年碩士論文

【摘要】：人體姿態(tài)識別是計算機視覺領(lǐng)域中最具挑戰(zhàn)性的研究方向,被廣泛應(yīng)用于人機交互、智能監(jiān)控、虛擬現(xiàn)實、人體行為分析等領(lǐng)域。隨著微軟公司推出Kinect深度傳感設(shè)備之后,基于深度圖像的姿態(tài)識別技術(shù)得到了越來越多的學者關(guān)注和研究。但是,在研究中由于圖像受光照、陰影外界變化以及人體姿態(tài)類別的限制、特征選擇的難度、機器學習的局限性、識別算法的性能等因素的影響,人體姿態(tài)識別研究進展緩慢,姿態(tài)識別的方法成為近些年的研究熱點。目前,人體姿態(tài)識別采用的方法大多是基于Kinect的骨架信息和深度圖像信息,通過對圖像分割實現(xiàn)人體前景提取,再利用HOG、HIFT或骨架關(guān)鍵坐標對人體姿態(tài)特征進行提取和降維,最后對模型進行訓練和分類。雖然這些方法在識別精確度和準確性方面效果比較好,但是由于人體復(fù)雜度較高,通過骨骼關(guān)鍵點和深度圖像信息提取的特征往往維度也比較高,對深度圖像進行特征提取前大多還需要對圖像進行一些預(yù)處理,導致特征提取難度大和識別效率不高。因此,本論文提出了一種基于卷積神經(jīng)網(wǎng)絡(luò)的人體姿態(tài)識別研究方法。主要研究內(nèi)容和工作任務(wù)是:(1)概述了深度圖像的成像原理和獲取方法,通過Kinect拍攝不同體型的人在不同背景、光照條件下的圖片,構(gòu)建了包含十個姿勢的訓練樣本以及測試樣本,以保證數(shù)據(jù)庫的多樣性。(2)介紹了人工神經(jīng)網(wǎng)絡(luò)和卷積神經(jīng)網(wǎng)絡(luò)的相關(guān)理論。重點介紹了經(jīng)典的前饋神經(jīng)網(wǎng)絡(luò)的結(jié)構(gòu)及反向傳播算法,以典型的卷積神經(jīng)網(wǎng)絡(luò)結(jié)構(gòu)為例,詳細闡述了網(wǎng)絡(luò)結(jié)構(gòu)的組成和訓練方式以及卷積神經(jīng)網(wǎng)絡(luò)相關(guān)知識如激活函數(shù)、損失函數(shù)、梯度計算、softmax回歸分類技術(shù)等。(3)針對卷積神經(jīng)網(wǎng)絡(luò)具有局部感知、權(quán)值共享、網(wǎng)絡(luò)模型的復(fù)雜度低、圖像可直接作為網(wǎng)絡(luò)的輸入以及對平移、比例縮放、傾斜等變形網(wǎng)絡(luò)結(jié)構(gòu)具有高度不變性等優(yōu)點,本文提出了基于卷積神經(jīng)網(wǎng)絡(luò)的人體姿態(tài)識別研究,并從網(wǎng)絡(luò)層數(shù)、濾波器個數(shù)、激活函數(shù)等方面優(yōu)化網(wǎng)絡(luò)結(jié)構(gòu),從而進行人體姿態(tài)的分類識別。
[Abstract]:Human posture recognition is the most challenging research field in the field of computer vision. It is widely used in human-computer interaction, intelligent monitoring, virtual reality, human behavior analysis and other fields.With the introduction of Kinect depth sensing device by Microsoft, the attitude recognition technology based on depth image has been paid more and more attention and research.However, due to the influence of illumination, shading outside changes and human pose types, the difficulty of feature selection, the limitation of machine learning, and the performance of recognition algorithm, the research progress of human pose recognition is slow.Attitude recognition has become a hot topic in recent years.At present, most of the human attitude recognition methods are based on the skeleton information and depth image information of Kinect. The human body foreground is extracted by image segmentation, and then the human pose feature is extracted and dimensionally reduced by using hog HIFT or skeleton key coordinates.Finally, the model is trained and classified.Although these methods have good recognition accuracy and accuracy, because of the high complexity of human body, the feature dimensions extracted from the key points and depth images of bone are often higher.Most of the depth images need to be preprocessed before feature extraction, which leads to the difficulty of feature extraction and the low efficiency of recognition.Therefore, this paper proposes a research method of human body attitude recognition based on convolution neural network.The main research content and work task is: (1) the imaging principle and acquisition method of depth image are summarized. The images of people of different body types under different background and illumination conditions are taken by Kinect.The training samples including ten postures and test samples are constructed to ensure the diversity of database. (2) the related theories of artificial neural network and convolutional neural network are introduced.The structure and back propagation algorithm of classical feedforward neural network are introduced in detail. Taking the typical convolutional neural network structure as an example, the composition and training mode of the network structure and the related knowledge of convolutional neural network such as activation function are described in detail.Loss function, gradient calculation and Softmax regression classification, etc.) the convolutional neural network has local perception, weight sharing, low complexity of the network model, and image can be directly used as the input of the network as well as translation and scaling.In this paper, the human body attitude recognition based on convolution neural network is proposed, and the network structure is optimized in terms of the number of network layers, the number of filters, the activation function, etc.Thus, the classification and recognition of human posture are carried out.
【學位授予單位】：成都理工大學
【學位級別】：碩士
【學位授予年份】：2017
【分類號】：TP391.41;TP183

【參考文獻】

相關(guān)期刊論文 前10條

1 嚴利民;李躍;;基于Kinect深度數(shù)據(jù)的人體骨架提取[J];電子測量技術(shù);2015年03期

2 陳曦;孟慶虎;;骨架關(guān)節(jié)點跟蹤的人體行為識別方法[J];河南科技大學學報(自然科學版);2015年02期

3 劉曉云;孫運強;姚愛琴;;基于骨架的人體識別[J];電子世界;2014年19期

4 劉飛;郝礦榮;丁永生;劉歡;;基于深度圖像的人體動作識別方法[J];計算機工程;2014年08期

5 李瑞峰;王亮亮;王珂;;人體動作行為識別研究綜述[J];模式識別與人工智能;2014年01期

6 王艷;張奇志;;基于Kinect深度信息的手勢識別[J];北京信息科技大學學報(自然科學版);2013年01期

7 黃國范;李亞;;人體動作姿態(tài)識別綜述[J];電腦知識與技術(shù);2013年01期

8 鄧瑞;周玲玲;應(yīng)忍冬;;基于Kinect深度信息的手勢提取與識別研究[J];計算機應(yīng)用研究;2013年04期

9 李紅波;丁林建;冉光勇;;基于Kinect深度圖像的人體識別分析[J];數(shù)字通信;2012年04期

10 林鵬;張超;李竹良;趙宇明;;基于深度圖像學習的人體部位識別[J];計算機工程;2012年16期

相關(guān)碩士學位論文 前10條

1 姚欽文;基于卷積神經(jīng)網(wǎng)絡(luò)的車臉識別研究[D];浙江大學;2016年

2 鄧柳;基于深度卷積神經(jīng)網(wǎng)絡(luò)的車型識別[D];西南交通大學;2015年

3 張騰飛;基于多特征融合的人體姿態(tài)識別[D];哈爾濱工業(yè)大學;2014年

4 劉飛;基于Kinect骨架信息的人體動作識別[D];東華大學;2014年

5 陳先昌;基于卷積神經(jīng)網(wǎng)絡(luò)的深度學習算法與應(yīng)用研究[D];浙江工商大學;2014年

6 黃莎;基于骨架的人體動作識別研究[D];上海交通大學;2014年

7 徐姍姍;卷積神經(jīng)網(wǎng)絡(luò)的研究與應(yīng)用[D];南京林業(yè)大學;2013年

8 殷海艷;基于深度圖像的人體姿態(tài)識別[D];北京工業(yè)大學;2013年

9 羅鳴;基于Kinect傳感器的骨骼定位研究[D];武漢科技大學;2013年

10 楊林;基于Kinect的人體目標檢測與跟蹤[D];大連海事大學;2013年

，

本文編號：1724344