English subtitles

← 17-33 Improving Stereo Vision

Get Embed Code
2 Languages

Showing Revision 1 created 11/28/2012 by Amara Bot.

  1. I'd like to say a few words about how to improve the results of stereo vision.
  2. Here is a vision assembly that James David built up of two cameras.
  3. In addition to having these two cameras, he also put a projector into the scene
  4. that emitted a random light pattern.
  5. In fact, it emitted a striped pattern, shown over here on this frog,
  6. and by adding texture to the scene, you can making correspondence easier.
  7. This is a striped pattern of unequal distances.
  8. There's a coding over here, which makes certain stripes larger than others.
  9. If you run the same algorithm I just told you,
  10. you'll find that stereo vision becomes better,
  11. because we can now better disambiguate the correspondence of points.
  12. Here is the assembly used for imaging myself. This is me with a sweater on.
  13. That's my face.
  14. And you can see by emitting structured light, as it is called,
  15. you can enhance the performance of stereo
  16. and objects that otherwise have very poor texture.
  17. Another solution is called the Microsoft Kinect. You're probably familiar with it.
  18. It's a new gaming platform that's been sold at record pace.
  19. It uses a camera system, together with a laser.
  20. The laser adds texture to the scene,
  21. and by triangulation using the same method I showed you,
  22. it can recover depth.
  23. Here's my postdoc Christian using a Kinect-like sensor
  24. to do certain poses in front of a depth sensor.
  25. You can see in the screen how his pose is being perceived,
  26. and you can see Christian trying to do handstands and other acrobatic maneuvers.
  27. He's actually pretty good.
  28. That's all using effectively stereo vision.
  29. There is actually a whole bunch of different types of techniques
  30. for sensing range in computer vision.
  31. I'm just going to briefly talk about them.
  32. They're called laser range finders.
  33. They send off beams of light,
  34. and they measure the time until the light comes back into the sensor.
  35. They're being manufactured by many different companies.
  36. In our experiments using robots to drive through the desert and through traffic.
  37. We quite extensively used laser range finders as an alternative to stereo vision,
  38. because they give us very, very good range estimates.
  39. Here is a 3D model constructed by laser range finders of our neighborhood in Palo Alto,
  40. and it's easy to see how 3D points can making amazing 3D models,
  41. using techniques like stereo vision or like the laser range finders I just briefly talked about.