How to convert Kinect raw depth info to meters in Matlab? - matlab

Question

I have made some research here to understand this topic but I have not achieved good results. I'm working with a Kinect for Windows and the Kinect SDK 1.7. I'm working with matlab to process raw depth map info.
First, I'm using this method (https://stackoverflow.com/a/11732251/3416588) to store Kinect raw depth data to a text file. I got a list with (480x640 = 307200) elements and data like this:
23048
23048
23048
-8
-8
-8
-8
-8
-8
-8
-8
6704
6720
6720
6720
6720
6720
6720
6720
6720
6736
6736
6736
6736
6752
0
0
Then in Matlab I convert this values to binary. So, I get 16-bits numbers. The last three numbers that correspond to player index are "000", so I remove them. Now, from the 13 bits numbers, I remove the more significative, which is always "0".
So, I have made this:
[0,1,1,0,0,0,1,0,0,0,1,1,1,0,0,0] - 16 bits number
[0,1,1,0,0,0,1,0,0,0,1,1,1] - 13 bits number
[1,1,0,0,0,1,0,0,0,1,1,1] - 12 bits number
Then I follow to this procedure (https://stackoverflow.com/a/9678900/3416588) to convert raw-depth info to meters but i get values in the range from -4.7422 to 0.3055. What do they mean?
clc
clear all
close all
%Get raw depth data from .txt
fileID = fopen('C:\Example.txt', 'r');
datos = fscanf(fileID, '%i'); % Data in decimal
fclose(fileID);
% If raw depth data is less than 0, set it to 0
for i = 1 : 640*480
if(datos(i) < 0)
datos(i) = 0;
end
end
% Auxiliar arrays
datosBin = zeros(640*480, 16);
realDepth = zeros(640*480, 12);
% Decimal raw depth data to binary
n = 16;
m = 0;
for i = 1 : 640*480
a = datos(i);
datosBin(i,:) = fix(rem(a*pow2(-(n-1):m),2));
end
% Remove player index and msb (more significant bit)
for i = 1 : 640*480
realDepth(i,:) = datosBin(i,2:13);
end
% Auxiliar array to store raw depth data decimal number
realDepthDec = zeros(640*480,1);
% Raw depth data 12 bits to decimal
for i = 1 : 640*480
realDepthDec(i) = bin2dec(num2str(realDepth(i,:)));
end
% Auxiliar array
rawDepthMapMeters = zeros(480, 640);
% Create array 640*480 to store bit depth info in meters
for j = 1 : 480
for i = 1 : 640
if(realDepthDec(i+j) <= 2046)
rawDepthMapMeters(j, i) = 0.1236 * tan(realDepthDec(i+j)/2842.5 + 1.1863);
end
end
end
Where is my mistake? What am I doing wrong? Thanks in advance.
PS. Sorry for my bad english.

Answer 1

In the second article you read, you will see that the method you use is outdated. Read this.
x = (i - w / 2) * (z + minDistance) * scaleFactor
y = (j - h / 2) * (z + minDistance) * scaleFactor
z = z
Where
minDistance = -10
scaleFactor = .0021.
These values were found by hand.
Also you could convert those points to millimeters in your first application as described in the second question
using (var depthFrame = e.OpenDepthImageFrame())
{
var depthArray = new short[depthFrame.PixelDataLength];
depthFrame.CopyPixelDataTo(depthArray);
for (int i = 0; i < depthArray.Length; i++) {
int depthInMillimeters =
depthArray[i] >> DepthImageFrame.PlayerIndexBitmaskWidth;
// TADAx2!
}