个性化虚拟手一体化建模平台的设计和开发

发布时间：2018-06-16 18:04

本文选题：个性化虚拟手 + 一体化平台　；参考：《浙江大学》2017年硕士论文

【摘要】：虚拟现实技术旨在高度模拟现实世界环境,让用户与虚拟环境交互,给人身临其境的感觉,它的两个重要特征是交互性和沉浸感。而手作为人与虚拟现实环境交互时的重要工具,逼真的手模型会让用户得到更加真实的体验。建立逼真的手模型面临许多挑战:自然环境下的很多因素会干扰人手图像采集,复杂背景导致人手与背景难以分离,光照环境影响人手的清晰度,人手的抖动会使人手图像模糊,软件环境需要过多的手动干预使建模周期长,效率低。本文开发了一体化的软硬件平台,使人手建模更加方便、快捷,建立的手模型更加真实。在硬件方面,设计并实现了人手多视图影像同步采集装置,该装置在考虑了人手生理特征的基础上,依据相机立体视锥、光照明模型、人手运动约束等原理进行了详细设计。使用我们的装置能减少人手图像采集过程中产生的环境噪音和运动噪音,提高采集到的图像质量。在软件方面,设计了面向用户的软件操作界面和高效实用的软件操作流程,使建模更加高效。根据计算机视觉中的多视图几何的原理,可以从人手的多视图影像中获取到三维信息,获取的信息分三类:反映人手大小、方位的点特征,反映人手轮廓的线特征,反映人手起伏的面特征。把人手模板模型按照点、线、面三种特征进行变形,获得大小、形状与真实手基本一致的手模型,最后将图片中真实手的纹理映射到虚拟手模型上,得到与用户手高度一致的逼真的手模型。将本文建立手模型的方法与三维激光扫描技术以及飞行时间技术建立的手模型进行比较发现:在操作上,本文的方法比后两者方便很多,在结果上比飞行时间技术建立的模型更具有真实感,同时本文用到的设备价格比三维扫描仪要便宜很多。
[Abstract]:Virtual reality technology aims to highly simulate the real world environment, make users interact with virtual environment, and give people a sense of reality. The two important characteristics of virtual reality technology are interactivity and immersion. Hand is an important tool for the interaction between human and virtual reality environment. The realistic hand model will make the user get more real experience. There are many challenges in creating realistic hand models: many factors in the natural environment interfere with human hand image collection, complex background makes it difficult to separate the hand from background, and illumination affects the clarity of the hand. The hand jitter will blur the hand image, and the software environment needs too much manual intervention to make the modeling cycle long and inefficient. In this paper, an integrated software and hardware platform is developed, which makes manual modeling more convenient, faster and more realistic. In the aspect of hardware, we design and implement the synchronous acquisition device of human multi-view image. Based on the physiological characteristics of human hand, the device is designed in detail according to the principles of camera stereoscopic cone, light illumination model, manual motion constraint and so on. The use of our device can reduce the environmental noise and motion noise generated in the process of manual image acquisition and improve the image quality. In the software aspect, the user oriented software operation interface and the efficient and practical software operation flow are designed to make the modeling more efficient. According to the principle of multi-view geometry in computer vision, 3D information can be obtained from multi-view images of human hands. The information obtained can be divided into three categories: points that reflect the size and orientation of the hands, line features of the contour of the hands. Reflects the face characteristics of the human hand undulation. The manual template model is deformed according to three features: point, line and surface, and the hand model with the same size and shape as the real hand is obtained. Finally, the texture of the real hand in the picture is mapped to the virtual hand model. A realistic hand model is obtained, which is highly consistent with the user's hand. Compared with the 3D laser scanning technique and the flying time technique, the method of establishing hand model in this paper is much more convenient than that of the latter. The result is more realistic than the model built by time of flight technology, and the price of the equipment used in this paper is much cheaper than that of 3D scanner.
【学位授予单位】：浙江大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.9

【参考文献】