Depth Image Capturing:
As we used depth camera as a tool, the data what we captured is so called “depth image”. It is a gray scaled image and the color of the pixel is determined by the distance between the shot and the object – the further distance results in shallower pixel color. So as you can see in the attached figure (Figure 2), the background is almost white because they are far away while the hand is gray because it is nearer.
Besides, making the captured data load to MATLAB is another problem. Here, we build a MEX file to link our C++ source of capturing the image with MATLAB, so the data can be directly saved in the form of matrix (as we used default settings, the matrix is 240*320 unit16). In order to show the image using MATLAB function “imread”, we normalize the data using following formula:
By then, the matrix “framed” consists of elements with its range from 0 to 1, which meets the function requirement. So the figure below (Figure 1) is what we want for the first step.
Figure 1: Original data
Figure 2: Depth image