体感技术：微软三维重建项目 KinectFusion

原文：http://user.qzone.qq.com/495830284/blog/1324983917

微软利用kinect深度信息进行三维重建，项目名字是kinectfusion。

下面是官网重建全过程的演示视频链接，只能说太强大了。像微软这样的大公司是不是有一条潜规则：我先内部做到一定高度，同时抛出一些演示视频和粗糙的技术（第一代kinect），让大家感兴趣（买产品），以此保持行业领先；类似与苹果的“饥饿”营销。

http://research.microsoft.com/en-us/projects/surfacerecon/

微软正在进行一个名为KinectFusion的项目。目标是利用一台围绕物体移动的kinect实时重建物体的三维模型。（是不是可以理解为 Kinect版本的SLAM？）不同于的简单的三维点云的拼接，该项目的另外的吸引人的特性在于：如果对物体进行持续的扫描，三维重建精度可以由粗到细的逐渐提高。

We present KinectFusion,a system that takes live depth data from a moving Kinectcamera and in real-time creates high-quality, geometrically accurate, 3Dmodels. Our system allows a user holding a Kinectcamera to move quickly within any indoor space, and rapidly scan and create afused 3D model of the whole room and its contents within seconds. Even smallmotions, caused for example by camera shake, lead to new viewpoints of thescene and thus refinements of the 3D model, similar to the effect of imagesuperresolution. As the camera is moved closer to objects in the scene moredetail can be added to the acquired 3D model.

To achieve this, our system continually tracks the 6DOFpose of the camera and rapidly builds a representation of the geometry ofarbitrary surfaces. Novel GPU-based implementations for both camera trackingand surface reconstruction allow us to run at interactive real-time rates thathave not previously been demonstrated. We define new instantiations of two wellknown graphics algorithms designed specifically for parallelizable GPGPUhardware.

下面是两篇技术论文，必然是微软的人写的：

作者 Shahram Izadi 的主页网址：http://research.microsoft.com/en-us/um/people/shahrami/

KinectFusion: Shahram Izadi, Richard Newcombe,David Kim, Otmar Hilliges, David Molyneaux, Steve Hodges, Pushmeet Kohli, JamieShotton, Andrew Davison, Andrew Fitzgibbon

“ KinectFusion: Real-Time Dynamic 3D SurfaceReconstruction and Interaction ”

ACM SIGGRAPH 2011 Talk
S. Izadi, D. Kim, O.Hilliges, D. Molyneaux, R. Newcombe, P. Kohli, J. Shotton, S. Hodges, D. Freeman,A. Davison, A. Fitzgibbon“KinectFusion: Real-time 3D Reconstruction andInteraction Using a Moving Depth Camera ”

To appear in Proceedings ACM User Interface andSoftware Technologies ( ACM UIST )

Santa Barbara, CA, USA, Oct 2011

下面是三维点云的开源库PCL简要介绍

目前还不够成熟，关键是“开源”两个字值钱。

http://pointclouds.org/

Point Cloud Library (PCL) 是一个独立的、大尺度的三维点云处理库

无论是商用还是研究都是免费的

Point cloud library 是一个开源、跨平台的点云处理库， pcl 框架涵盖了当前最先进的点云处理相关算法，涉及滤波、特征估计、表面重建、注册、模型拟合、分割 等。 Pcl 基于 BSD 协议，可免费用于商业和研究应用，适用于 Linux 、 MacOS 、 Windows 、 Android 平台。

赞助商有Willow Garage, NVidia, Google, Toyota, 以及 Trimble 

相关博客 http://qing.weibo.com/1455173150/56bc2e1e330005pk.html