發(fā)布時間：2018-04-18 08:18

  本文選題：運動捕捉技術(shù) + 數(shù)據(jù)處理　； 參考：《北方工業(yè)大學》2017年碩士論文

【摘要】：隨著虛擬現(xiàn)實技術(shù)的發(fā)展,虛實交互技術(shù)逐漸成為研究熱點和難點。虛實交互是將計算機生成的虛擬場景與真實世界場景相互融合的一種技術(shù),它是獲取真實世界中物理對象的運動軌跡,實現(xiàn)真實場景和虛擬場景之間的自然交互,呈現(xiàn)給用戶一種全新的人機交互體驗環(huán)境。傳統(tǒng)的虛實交互技術(shù)應(yīng)用局限較多,體驗感差。本文主要研究基于動捕信息的虛實交互技術(shù),最終設(shè)計并實現(xiàn)了一個沉浸感良好和實時效率較高的虛實交互系統(tǒng)。本文的主要研究成果如下:(1)針對人體實時交互過程中標記點被干擾產(chǎn)生噪聲數(shù)據(jù)的問題,研究并設(shè)計了基于卡爾曼濾波算法的標記點位置預測方法。人體作為一種新型的數(shù)據(jù)輸入方法,通過運動捕捉設(shè)備采集人體運動數(shù)據(jù)時,身上粘貼的標記點間可能存在相互干擾,要對標記點位置進行優(yōu)化。首先,分析人體骨骼結(jié)構(gòu),完成人體簡單幾何模型的構(gòu)建;然后,根據(jù)人體的運動特點,構(gòu)建出人體的層次結(jié)構(gòu)模型;最終,基于人體的層次結(jié)構(gòu)采用卡爾曼濾波算法來估計下一幀標記點的方位信息對數(shù)據(jù)進行處理,提高標記點跟蹤的準確性。實驗驗證表明,本文的基于卡爾曼濾波的標記點位置預測方法使數(shù)據(jù)更加平滑。(2)針對虛擬場景中交互的穿越問題,研究并設(shè)計了基于層次包圍盒的碰撞檢測方法。首先,對包圍盒分層構(gòu)建和遍歷,用該包圍盒替代虛擬對象進而判斷物體間是否發(fā)生碰撞;然后,基于該碰撞檢測算法引入Unity3d游戲引擎開發(fā)的多種碰撞器,并根據(jù)虛擬對象的幾何特征組合碰撞器,對于發(fā)生碰撞的物體及時做出響應(yīng)。實驗驗證表明,本文研究的基于層次包圍盒的碰撞檢測方法有效地提高了碰撞檢測的效率,并防止發(fā)生穿越。(3)為了更好的應(yīng)用以上兩個方法,設(shè)計并實現(xiàn)了一個虛實交互系統(tǒng)。本文為了實現(xiàn)參與者對虛擬對象的動作、位置、朝向、姿態(tài)的控制,劃分了幾何模型構(gòu)建模塊、數(shù)據(jù)處理模塊、碰撞檢測模塊、虛實交互模塊,最終完成了系統(tǒng)。系統(tǒng)運行效果表明,基于動捕信息的虛實交互系統(tǒng)能夠較好的實現(xiàn)人與剛體、剛體與剛體、人與純虛模型的多種虛實和虛虛交互。
[Abstract]:With the development of virtual reality technology, virtual reality interaction technology has gradually become a hot and difficult point.Virtual interaction is a kind of technology that combines the computer generated virtual scene with the real world scene. It is to obtain the movement track of the physical object in the real world and to realize the natural interaction between the real scene and the virtual scene.Presents the user a kind of brand-new human-machine interaction experience environment.The traditional virtual reality interaction technology has many limitations and poor experience.This paper mainly studies the virtual and real interaction technology based on moving capture information, and finally designs and implements a virtual reality interaction system with good immersion and high real time efficiency.The main research results of this paper are as follows: (1) aiming at the problem of the noise data caused by the interference of the mark points in the human real-time interaction process, this paper studies and designs a method for prediction of the location of the mark points based on Kalman filtering algorithm.As a new data input method, the human body may interfere with each other when collecting human motion data by motion capture equipment, so the location of labeled points should be optimized.First, analyze the human skeleton structure, complete the construction of the simple geometric model of the human body; then, according to the characteristics of human movement, construct the hierarchical structure model of the human body; finally,Based on the hierarchical structure of human body, Kalman filter algorithm is used to estimate the orientation information of the next marker point to process the data to improve the accuracy of the tracking of the mark point.The experimental results show that the method based on Kalman filter makes the data smoother. (2) aiming at the problem of interaction traversing in virtual scene, the collision detection method based on hierarchical bounding box is studied and designed.Firstly, the bounding box is constructed and traversed in layers, the bounding box is used to replace the virtual object to judge whether there is collision between objects, and then, based on the collision detection algorithm, a variety of colliders developed by Unity3d game engine are introduced.According to the geometric features of virtual object, the collider can respond to the colliding object in time.Experimental results show that the proposed collision detection method based on hierarchical bounding box can effectively improve the efficiency of collision detection and prevent the collision detection from occurring. (3) in order to better apply the above two methods, a virtual and real interactive system is designed and implemented.In order to control the action, position, orientation and attitude of the virtual object, this paper divides the geometric model building module, the data processing module, the collision detection module, the virtual interaction module, and finally completes the system.The results show that the virtual interaction system based on moving capture information can realize the interaction between human and rigid body, rigid body and rigid body, and virtual interaction between human and pure virtual model.
【學位授予單位】：北方工業(yè)大學
【學位級別】：碩士
【學位授予年份】：2017
【分類號】：TP391.9

【參考文獻】

相關(guān)期刊論文 前10條

1 侯永隆;寧濤;王可;;基于光學運動捕捉的虛擬人體標定技術(shù)[J];圖學學報;2013年05期

2 邱世廣;周德吉;范秀敏;武殿梁;;虛擬操作仿真環(huán)境中基于運動捕獲的虛擬人實時控制技術(shù)[J];計算機集成制造系統(tǒng);2013年03期

3 王軍;胡永剛;韓崇昭;;被動式光學運動捕捉系統(tǒng)丟點檢測與補償[J];系統(tǒng)工程與電子技術(shù);2012年11期

4 張毅;張爍;羅元;徐曉東;;基于Kinect深度圖像信息的手勢軌跡識別及應(yīng)用[J];計算機應(yīng)用研究;2012年09期

5 魏小鵬;劉瑞;張強;肖伯祥;;基于模板匹配的人體運動捕捉數(shù)據(jù)處理方法[J];系統(tǒng)仿真學報;2010年10期

6 吳升;張強;肖伯祥;魏小鵬;;一種有效的光學運動捕捉數(shù)據(jù)處理方法[J];應(yīng)用基礎(chǔ)與工程科學學報;2009年05期

7 王偉;馬峻;劉偉;;基于OBB包圍盒的碰撞檢測研究與應(yīng)用[J];計算機仿真;2009年09期

8 肖伯祥;魏小鵬;張強;周昌軍;;光學運動捕捉散亂數(shù)據(jù)處理的一種方法[J];系統(tǒng)仿真學報;2008年02期

9 馬登武;葉文;李瑛;;基于包圍盒的碰撞檢測算法綜述[J];系統(tǒng)仿真學報;2006年04期

10 陳學文,丑武勝,劉靜華,王田苗;基于包圍盒的碰撞檢測算法研究[J];計算機工程與應(yīng)用;2005年05期

相關(guān)博士學位論文 前1條

1 肖伯祥;運動捕捉數(shù)據(jù)處理、檢索與重構(gòu)方法研究[D];大連理工大學;2009年

相關(guān)碩士學位論文 前1條

1 周曉陽;基于光學運動捕捉的虛擬交互系統(tǒng)開發(fā)[D];哈爾濱工業(yè)大學;2012年

，

本文編號：1767542