VolVision: High-speed Capture in Unconstrained Camera Motion

彬彬

Abstract

In this paper, we propose a novel concept called VolVision that en-

compasses using a camera to recons***ct 6DoF unconstrained mo-

tion. ”VolVision” is designed to handle imagery falling, tossed or

thrown cameras. And VolVision also allows users to recons***ct

dynamic images and generate a 3D-mapped scene from image se-

quences. It could be used to model severe environments like valley

and mountains that are normally not easily viewed by humans. We

produced a prototype that embodies the concept above, and were

able to recons***ct the camera’s path, perform image mosaicing,

and track 3D information of feature points in images.

CR Categories: I.4.8 [Image Processing and Computer Vision]:

Scene Analysis—Motion;

Keywords: high-speed vision, 6DoF motion, tracking

1 Introduction

Late model digital cameras are primarily distinguished from prede-

cessors by the small form-factor which has allowed their incorpo-

ration into many mobile devices. Correspondingly, a new crop of

research has started to examine what can be done with pervasive

mobile cameras. For example, Phototourism [Snavely et al. 2006]

allows users to travel virtually in a 3D space by inferring each cam-

era’s position and orientation and rendering a scene recons***ction.

Such developments in the camera modules and usage open the

possibility of more compelling experiences than just taking pho-

tos. However, in terms of camerawork, those applications always

require fixed or stable motion for getting clear, blurless images.

For example, photographers often limit the shooting range during

capture through implements like tripods, rails, and cranes. Such

constraints along with special stabilization technologies such as

Steadicam or Sony SteadyShot are the most popular ways to get

stable images.

In video imagery, professional and amateur directors alike affix

cameras to cars, airplanes that are human-controlled. Directors of-

ten limit the shot range or prepare overly complex capturing sys-

tems in order to reduce the dimensionality of camera motion and

get a stable image.

On the other hand, there is some research into 6DoF motion cam-

eras, but most of this concerns handheld camera system, so those

system assume that camera motion is slower than motions like

falling, tossing, and throwing. It is frequently assumed that pho-

tography can’t take place during such moments, but many of the

most compelling evidentiary images involve fast-moving and unex-

pected shifts in camera pose (for instance onboard vehicles cameras

C.R.CAN

鉴于你今年的良好表现，新年将至，别人都在祝福你新年快乐，我觉得换一种新颖的方式祝福你，那就是：天天开心，事事顺利，雷打不动的祝福属于你！

菜刀吻电线

快绿怡红旦暮有，乐得共酒共心欢.

晃晃

真不错,全存下来了.

菜刀吻电线

百度的叫度娘，网易的叫易娘，新浪内部还在为是叫新娘还是浪娘而争论不休！……不管你们是企鹅的额娘，豆瓣的伴娘，还是华为的伪娘，都要记得，淘宝才是你们的亲娘啊！亲！！

菜刀吻电线

呵呵,真得不错哦!!