RGBDCapture

This code is to capture RGB-D data into PNG images with an OpenNI-drived depth camera (Kinect v1, Asus XTion, etc). The RGB images are 640x480 8-bit images, and the depth images are 640×480 16-bit monochrome images. The RGB-D data is exactly in the format of TUM RGB-D data. You can use their tools to analyze the data.

This code is written in C++ and is tested successfully in Visual Studio 2013 in Windows and Ubuntu 14.04.

An example of scanned images using Kinect v1:

Dependencies

• OpenCV 2.4.X -- images processing
• OpenNI2 -- driver
• Boost (>= 1.50) -- file systems

Build

To compile the code:

• In Windows, use Visual Studio to open .sln file and compile.
• In Linux, use the standard approach

  $ mkdir build
  $ cd build
  $ cmake ..
  $ make

Note to modify corresponding paths of dependencies.

Usage

./RGBDCapture

No input options.

Note

• Press ESC to quit scanning.
• The RGB-D data will be saved in a newly created folder named saveX, where X is an index integer starting from 0.
• Check the path of OPENNI2 driver files if the code cannot start.
• You can use another code of mine RGBDConverter to convert RGB-D images into a single compressed file in KLG format, which can be taken as input of ElasticFusion code.
• This code doesn't give camera intrinsic parameters.