Supporting material for the article "Correction and Interpolation of Depth Maps from Structured Light Infrared Sensors"

If you are using materials from this page (software, datasets, etc.) please quote the paper:

Simone Milani and Giancarlo Calvagno. 2016. Correction and interpolation of depth maps from structured light infrared sensors. Image Commun. 41, C (February 2016), 28-39. DOI=http://dx.doi.org/10.1016/j.image.2015.11.008 pdf.


Source code: denCode.zip.

We report here the dataset employed to generate the experimental results reported in the paper.

Dataset consists in two folder related to two different 3D scene:










The scene has been acquired using a NextEngine 2020i Desktop Laser scan. The generated 3D point cloud is present in each folder in Polygon File Format or the Stanford Triangle Format. The file name is ref_ply.ply Each scene has been acquired under different illumination conditions, which are associated to the relative subfolders:

Additional acquisitions were performed introducing a second Kinect device in the environment, pointed towards the scene. The different acquisitions were obtained changing the angle with respect to the reference Kinect device in order to vary the level of cross-talk noise between the two devices.These acquisitions are reported in the subfolders angle, where letter X correspond to different angles a according to the following scheme:
A B C D E
90° 54° 36° 18°


Within each folder, it is possible to find 10 different acquisitions consisting in couples of files related to color and depth information. Files are named as

Each subfolder is zipped into the archive .zip in order to allow an easier download.
The calibration parameters for the kinect device are available here.
For calibration, warping, etc.., we adopted the software Kinect Calibration Toolbox
As for ICP, we employed the implementation of Per Bergstrom available here.