A multiple camera with real time volume reconstruction for articulated skeleton pose tracking
This presentation is the property of its rightful owner.
Sponsored Links
1 / 19

A Multiple Camera with Real-Time Volume Reconstruction for Articulated Skeleton Pose Tracking PowerPoint PPT Presentation


  • 88 Views
  • Uploaded on
  • Presentation posted in: General

A Multiple Camera with Real-Time Volume Reconstruction for Articulated Skeleton Pose Tracking. 指導教授:王聖智 教授 學生:謝佳峻. Zheng Zhang, Hock Soon Seah1 Chee Kwang Quah,Alex Ong , and Khalid Jabbar

Download Presentation

A Multiple Camera with Real-Time Volume Reconstruction for Articulated Skeleton Pose Tracking

An Image/Link below is provided (as is) to download presentation

Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author.While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server.


- - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - -

Presentation Transcript


A multiple camera with real time volume reconstruction for articulated skeleton pose tracking

A Multiple Camera with Real-Time Volume Reconstruction for Articulated Skeleton Pose Tracking

指導教授:王聖智 教授

學生:謝佳峻

Zheng Zhang, Hock Soon Seah1 CheeKwangQuah,AlexOng, and Khalid Jabbar

K.-T. Lee et al. (Eds.): MMM 2011, Part I, LNCS 6523, pp. 182–192, 2011.Springer-VerlagBerlin Heidelberg 2011


Outline

Outline

  • Introduction

  • Multi-camera System

  • Volume Reconstruction

  • Skeleton Pose Estimation

  • Results

  • Conclusion


Outline1

Outline

  • Introduction

  • Multi-camera System

  • Volume Reconstruction

  • Skeleton Pose Estimation

  • Results

  • Conclusion


Introduction

Introduction

  • Markerless

    don’t need markers or special suits.

  • Multi-view

    deal better with occlusion and appearance ambiguity problems.

剪出主要物件

還原個體輪廓形狀

偵測動作與行為

建立場景資訊


Outline2

Outline

  • Introduction

  • Multi-camera System

  • Volume Reconstruction

  • Skeleton Pose Estimation

  • Results

  • Conclusion


Multi camera system

Multi-camera System

  • System Setup

    1.Cameras work synchronously for acquiring multiple image in time.

    2. The frame rate of image acquisition should be at least 15 fps.

    3. The bandwidth is sufficient for supporting the transfer of multi-video streams.

    4.The acquisition room ought to be large.

Only one PC !!


Outline3

Outline

  • Introduction

  • Multi-camera System

  • Volume Reconstruction

  • Skeleton Pose Estimation

  • Results

  • Conclusion


Volume reconstruction

Volume Reconstruction

  • Background Subtraction

Background modeling constructs a reference image representing the background.

Threshold selection determines appropriate threshold values used in the subtraction operation.

Subtraction operation or pixel classicationclassies the type of a given pixel, i.e., the pixel is the part of background, or it is a moving object.

: 目前影像

: 參考背景

: 為一門檻值


Volume reconstruction1

Volume Reconstruction

  • Shape-from-Silhouette and Visual Hulls

1.Each multi-view silhouette contour is firstly obtained.

2.Silhouette polygons are back-projected into their corresponding camera positions.

3. Volume reconstruction method

4.Testing each voxel’s 6-connected neighbors.


A multiple camera with real time volume reconstruction for articulated skeleton pose tracking

voxel

texture

(a)

(b)

(c)

Illustration of volume reconstruction rendered in point clouds (a), voxels without texturing (b) and voxels with texturing (c)


Outline4

Outline

  • Introduction

  • Multi-camera System

  • Volume Reconstruction

  • Skeleton Pose Estimation

  • Results

  • Conclusion


Skeleton pose estimation

Skeleton Pose Estimation

  • The body model

  • Barrel model

  • 10 body segments

29 DOFs

(1)

(2)


Skeleton pose estimation1

Skeleton Pose Estimation

  • PSO(particle swarm optimization)

is the position of the i-th particle at k-th iteration .

is the velocity of the i-th particle at k-th iteration .

represents a vector of random numbers uniformly distributed in

is the history best position found by the i-th particle.

is the global best position found by its neighborhood so far.

is a constriction coefficient .


Outline5

Outline

  • Introduction

  • Multi-camera System

  • Volume Reconstruction

  • Skeleton Pose Estimation

  • Results

  • Conclusion


Results

Results


Results1

Results


Outline6

Outline

  • Introduction

  • Multi-camera System

  • Volume Reconstruction

  • Skeleton Pose Estimation

  • Results

  • Conclusion


Conclusion

Conclusion

1.Real-time volume sequences are reconstructed for articulated pose recovery.

2.Relies on single PC.

3.Different body segments are not allowed to intersect in the space .

4.Different model points should avoid taking the same closest feature point.

Future work will concentrate on enhancing the tracking robustness and accurateness.


References

References

  • Horprasert, T., Harwood, D., Davis, L.S.: A statistical approach for real-time robust background subtraction and shadow detection.

  • Laurentini, A.: The visual hull concept for silhouette-based image understanding .

  • Matusik, W., Buehler, C., McMillan, L.: Polyhedral visual hulls for real-time rendering.

  • Clerc, M., Kennedy, J.: The particle swarm - explosion, stability, and convergence in a multidimensional complex space.

  • http://www.csie.ntu.edu.tw/~cyy/courses/vfx/05spring/lectures/scribe/12scribe.pdf


  • Login