虚拟场景中基于Motion Vector的人群图像绘制算法  

People-crowd rendering inside virtual scene by image-based rendering algorithm using Motion Vector

作　　者：霍奕[1,2] 王喜年[1] 张瑜[3] 陈敬利[4] 马海滨[1] 张鹏飞[5] 

机构地区：[1]河北师范大学职业技术学院,石家庄050031 [2]北京工业大学计算机学院,北京100022 [3]河北科技大学高教研究所,石家庄050018 [4]河北师范大学物理科学与信息工程学院,石家庄050031 [5]石家庄市公安局,石家庄050091

出　　处：《计算机工程与应用》2009年第23期173-175,199,共4页Computer Engineering and Applications

基　　金：北京市科委项目(No.D0605002040121);河北师范大学青年基金(No.L2007Q27)

摘　　要：针对虚拟场景中大规模人群图像绘制时由于采集图像数量有限,在视点连续变化时出现跳变的问题,提出了一种基于Motion Vector的图像绘制算法,该算法通过移动原始图像的像素位置生成相邻视点位置的图像,使得各个离散采样之间能够连续过渡,很好地解决了由于采样不足而引起的跳变问题,同时保证了绘制系统的实时性,实验结果验证了该算法的有效性。For image-based rendering method toward people crow rendering inside virtual scene,since the sampled images are limited,when viewpoint changes continuously in rendering process,there would appear leap problem.This paper proposes an image based rendering algorithm that uses Motion Vector to solve the problem,which produces new images for adjacent viewpoints by moving each pixel＇s position from the original images,achieves smooth transition between the discrete sampled images,solves the leap problem resuhed from the limited sampled images,and increases rendering speed at the same time.Experiment shows that this method is valid.

关 键 词：虚拟现实 人群仿真 基于图像的绘制 运动矢量 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]