1 / 14

3-D Object Recognition From Shape

3-D Object Recognition From Shape. Salvador Ruiz Correa Department of Electrical Engineering. The Problem. 3-D Model (database). 3-D Scene. Difficult Problem. Cloud of points. 3-D scanner. Scene. Pose Estimation. Recognition. Recovering. Recognition. Scene. Model. C. A. C’. B’.

cbryant
Download Presentation

3-D Object Recognition From Shape

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. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. 3-D Object Recognition From Shape Salvador Ruiz Correa Department of Electrical Engineering

  2. The Problem 3-D Model (database) 3-D Scene

  3. Difficult Problem Cloud of points 3-D scanner Scene Pose Estimation Recognition Recovering

  4. Recognition Scene Model C A C’ B’ B A’ Surface Matching using local “features”

  5. Surface Matching • Search for candidate ‘point correspondences’ between the • the model and scene that are geometrically consistent. • (use distance measure in the “space of local features”). • Generate a set of potential matches (groups of 3 or more • correspondences). • Keep the match that produces the best alignment between • Surface meshes (rotation matrix and translation).

  6. Local Features

  7. Spin Signature Generation Local descriptors of shape Pose invariant Robust to clutter and occlusion

  8. Spherical Spin Images (SSI)

  9. Recognition using SSI

  10. Qualitative Results Scenes Recognition

  11. Results % Improvement • Tested four methods ~ 4000 experiments • Standard Spin Images vs Spherical Spin Images. • (SI vs SSI) • SI with PCA vs SSI with Random Projections • (SIC + SSIC)

  12. Current Research • Recognition of well defined classes of objects based on 3-D shape. • Use techniques from statistical learning theory (support vector machines) and knowledge from the physiology of learning in humans to achieve categorization, detection and recognition of shape classes.

More Related