双人投射空间增强现实：Dyadic AR系统

需积分: 9 34 浏览量更新于2024-09-09 收藏 3.3MB PDF 举报

"Dyadic AR_UIST2014" 本文介绍的是"Dyadic Projected Spatial Augmented Reality"，这是在UIST（User Interface Software and Technology）2014会议上发表的一项研究，由Hrvoje Benko, Andrew D. Wilson和Federico Zannier等人共同完成，他们都是微软研究院的成员。该研究关注的是双人交互的增强现实（AR）技术，旨在提升两个人面对面交流时与共享虚拟场景互动的能力。在传统的增强现实系统中，用户通常通过手持设备或透视头戴显示器来与虚拟世界交互。然而，"Dyadic AR"系统引入了一种新的方法，它结合了动态投影映射、多个视角视图以及无设备交互，使得两个用户可以在面对面的设置中直接与3D虚拟对象进行互动，而不依赖于特定的硬件设备。论文中的关键概念"dyadic projected spatial augmented reality"（双人投射空间增强现实）是指将投影技术应用于实际环境，使两个人能够看到并交互同一个虚拟场景。如图1所示，系统会考虑房间的几何结构、用户的相对位置以及投影仪和摄像头的配对。每个用户的目标视图会在屏幕外渲染，然后投影的图像会根据表面几何形状（包括另一用户的位置）进行扭曲校正，以确保每个用户看到的虚拟场景与其期望的视图相匹配。这一创新的AR系统的主要优势在于，它克服了手持设备或头戴显示器的限制，增强了人际交流的自然性和沉浸感。通过这种方式，用户可以直接用手与虚拟物体交互，而无需手持设备，这增加了交互的直观性和物理反馈。此外，由于双方都能看到对方与虚拟世界的互动，这种共享的体验可以促进合作和沟通。该系统的设计考虑了多人协作的场景，比如工作讨论、教育演示或游戏，它提供了一种全新的、增强的社交互动方式。未来的研究可能会进一步探索如何优化这种技术，以提高用户体验，例如通过更精确的追踪和更逼真的视觉效果，以及在不同环境和用户需求下的适应性。 "Dyadic AR"是一种有潜力改变我们与虚拟世界交互方式的技术，特别是在促进面对面交流和协作的环境中。它不仅展示了增强现实技术的创新应用，还强调了人机交互设计在增进人际交流中的重要性。

Immersive Displays with Multiple Perspective Views

Providing multiple people with personal perspective views

in augmented or virtual environments has traditionally been

achieved through head-worn displays [7], where each user

has their own view on their own personal display. In practice,

the obstruction of the face with such displays and their

narrow field of view can lead to unnatural face-to-face

interactions. Our solution avoids this difficulty.

Alternatively, projection-bases systems rely on either time-

or space-multiplexing the projected images to show multiple

independent views. Agrawala et al. [2] demonstrates a time-

multiplexed approach with synchronized shutter glasses to

enable two stereoscopic user perspectives on an interactive

tabletop. While conceptually simple, this approach tends to

suffer from low image brightness and sometimes perceivable

flicker since each eye only gets a small slice of the available

light in each frame.

In contrast, Bimber et al.’s Virtual Showcase [5] uses multi-

plane beam combiners to enable up to four independent

perspectives on a spatially-multiplexed tabletop: each view

appears at a different location of the display and is optically

combined with four mirrors. Similarly, IllusionHole [21]

uses spatial-multiplexing to provide multiple participants

correct perspective views around the tabletop. Their display

greatly reduces the visible display area for each user to

ensure minimal image overlap. Our work also employs

spatial multiplexing approach to support multiple views. In

contrast, we exploit the fact that in the face-to-face

arrangement, most of the surfaces that each person sees are

on their partner’s body or in their partner’s background.

These surfaces are good for projections of individual views,

as they are not easily perceived by their partner.

Depth Perception and Object Presence

The ability of our system to provide two users with a sense

of spatial presence in a room-scale augmented reality hinges

on the human ability to perceive depth from a perspective

view without binocular cues. Conventional measures of

person- and object-presence have mostly been defined for

virtual environment displays that surround and isolate the

user [31]. Stevens and colleagues found that the users can

experience a measurable sense of object-presence with

projection-augmented models [33]. However, their

questionnaire-based study examined only planar projections

and not perspective views.

The relative importance of various depth cues in perception

of virtual objects [9] is an important consideration for our

system since we do not offer stereoscopic vision.

Sollenberger and Milgram [32] showed a large improvement

of head-coupled stereo over static non-head-coupled non-

stereo displays while Arthur et al.’s experiments [1] showed

that users greatly preferred head coupling without stereo to

stereo head-coupled displays in fish-tank VR. How such

results apply to perspective SAR configurations remains

unclear. The most closely related work in this space is the

depth perception study of MirageTable [3] which showed

substantial accuracy in users’ estimates of depth on a SAR

tabletop with head-coupled stereo view. Also related is a

pilot experiment reported by Broecker et al. [6] who

investigated a variety of cues affecting depth perception in a

view-dependent near-field SAR, but found no statistically

significant results.

While focusing on projected tabletops, Hancock et al. [11]

evaluated people’s ability to judge object orientation under

different projection conditions. Their work highlights the

importance of correct perspective on judgment of objects’

spatial presence especially in multi-user scenarios. There is

also a long line of related research in cognitive psychology

on understanding the relative importance of different cues for

depth perception (e.g., [9, 10, 30]). The complete review of

this work is beyond the scope of this paper, but the three

volume book by Howard [16] offers the definitive summary

of knowledge on the topic.

SYSTEM DESCRIPTION

We describe our prototype dyadic SAR system, including

hardware configuration, scene modeling, dynamic projection

mapping and support for multiple simultaneous views.

Technical contributions of our work include: a particular

configuration of pro-cam units to support dyadic interaction,

a graphics pipeline that blends views from two simultaneous

perspectives while supporting projection onto dynamic depth

maps (e.g., people), and the design of interactions and

experiences that showcase these capabilities.

Hardware Configuration

Our prototype dyadic SAR system employs three HD video

projectors (BenQ W1080ST), each paired with a Kinect for

Windows v2 sensor. Their mounting was chosen to both

display and sense around two users that are approximately

facing each other in a large room. Two of the projector and

camera pairs are mounted on the ceiling, about two feet

above the heads of each of the two users. These are oriented

so that they approximately face one another, covering the

opposite walls and part of the floor (Figure 1). Roughly

speaking, each user’s view is rendered by the projector above

them. The precise surface geometry necessary for dynamic

projection mapping for a user’s view is provided by the

Kinect paired with the projector above them. Meanwhile,

body tracking of that user is supported by the opposite facing

Kinect camera. While mounted significantly higher than in

most applications, Kinect body tracking works well in this

configuration. This symmetric arrangement of projectors and

cameras follows the symmetric nature of dyadic interaction.

The third projector and camera pair is mounted on the

ceiling, facing downwards, to cover the area between the

areas covered by the first two projectors.

Our current implementation is primarily hosted on a single

PC which drives all three projectors. As the current Kinect

for Windows v2 SDK can support only one camera per PC,

we have three additional PCs which send the Kinect depth

images and other image processing results (e.g., body

tracking, optical flow) to the main PC via network. All depth

剩余10页未读，继续阅读

wangwenfeng1

粉丝: 1
资源: 4

双人投射空间增强现实：Dyadic AR系统

Dyadic_Interaction_Analysis_for_Medical_Data_in_DC_VA_MD:计算DC-VA-MD区域中从事手术的医生之间的双向相互作用

sequency, normal Hadamard, and dyadic matlab程序

Dyadic Green Functions in Electromagnetic Theroy

Dyadic Green函数用于均匀化超颖表面的偶极激发

MayaTSubdiv:Maya-Plugin以及Siggraph 2015论文“ Dyadic T-mesh细分”的示例实现

dyadic wavelet transform.m，二进小波变换在图像压缩和增强中的应用。

半空间 Sommerfeld 积分器电场：该程序使用 Dyadic Green 函数来计算由顶部介质电偶极子或磁偶极子引起的精确场。-matlab开发

dynamic_network_embedding

eetop.cn_PowerSI_3DBEM_Tutorial

ucinet.rar_UCINET_Ucinet-for-windows

最新资源