RGB to Depth conversion

Basic information

Depth information captured from Kinect is stored in 16-bit data structure as shown below:

D13

D12

D11

D10

D9

D8

D7

D6

D5

D4

D3

D2

D1

U3

U2

U1

Real depth information  (13 bits)

User indicators (3 bits for 7 users)

User indicators: 3 lower value bits that indicate the users

Real depth information:13 lower bit indicate the distance between detected object and the Kinect depth sensor. 

The accuracy of the IR sensor is 7 mm, therefore the first 3 lower bit are always zero

0

0

0

D13

D12

D11

D10

D9

D8

D7

D6

D5

D4

0

0

0

Real depth information (13 bits)

RGB to depth conversion

In the sample video, the depth value is converted to 24-bit RGB value using Nui_ShortToQuad_Depth() from Kinect SDK, the following equations are used:

BYTE red= s;

BYTE green= s>>3 & 224;

BYTE blue3= s>>5 & 192;

Blue								Green								Red
D13	D12	0	0	0	0	0	0	D11	D10	D9	0	0	0	0	0	D8	D7	D6	D5	D4	0	0	0

In order to extract real value from the RGB representation, we need to do following mappings:

$Depth\;in\;mm=2^5 blue + 2^3 green + red$