How Kinect works with PrimeSense -2

Supplement for my article "How Kinect works with Prime Sense"
Here lists the reference patents of the diagram and pictures.

• Method and system for object reconstruction
  (WO2007/043036)
• Distance-varying illumination and imaging techniques for depth mapping (US2010/0290698)
• Three-dimensional using speckle patterns
  (WO2007/105205)

I almost read all of PrimeSense's patterns, no matter of USPTO or WIPO. From these patterns I guess how kinect works and admire they use such complicated techniques to get such z-depth at low cost (even at low cost in calculation).

Besides the pseudo random pattern shows in the patent

• Depth-varying light fields for three dimensional sensing (US2008/0106746)

The two pictures show at different z-position. Pattern from superposition of diffuser and DOE is projected as pseudo random type, which is important to speckle in this application. Speckle autocorrection is the basic principle for the triangulation, but constricts the sensing range. So the coherent properties can't too much or too less, quasi- pseudo speckle pattern is suitable for calculating each spot in local z-variation and the absolute position information is from the shape properties of each spot. From the embodiment example form the pattern, the picture has 200x200 speckle spots in this 1000x1000 pixels picture.

If you has any questions or correction for my article, welcome give comments on my blog.

[Related article]
How Kinect works with PrimeSense