Supporting material for the article "Joint Denoising and Interpolation of Depth Maps for MS Kinect Sensors"

The current page reports extra material, datasets, and details about the paper


S. Milani, G. Calvagno, "Joint Denoising and Interpolation of Depth Maps for MS Kinect Sensors", Proc. of ICASSP 2012, Kyoto, Japan, Mar. 25-30, 2012. Bibtex PDF Video demo available here.

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

The test set consists in the folders

• bearball (zipped version here)
• duckwater (zipped version here)
• duck (zipped version here)
• bins (zipped version here)
• bike (zipped version here)
• coffee (zipped version here)
• computer (zipped version here)
• director (zipped version here)
• entering (zipped version here)
• parking (zipped version here)
• peter (zipped version here)
• mail (zipped version here)
• soda (zipped version here)
• paper (zipped version here)

Each folder contains images related to 10 independent acquisitions. Each folder is compressed in the corresponding zip file. The directory results_for_denoising contains some denoised data (using our denoising algorithm) The acquisition were made under different lights and environments. bearball, duckwater, duck: controlled-diffused light/indoor/no reflections bins: uncontrolled sun light/indoor (corridor)/many windows and reflections paper: uncontrolled neon light/indoor (corridor)/no windows mail: uncontrolled sun light/indoor (corridor)/medium reflections director: poor uncontrolled lighting/indoor (corridor)/medium reflections soda: uncontrolled neon and sun light/indoor/close to window computer: uncontrolled neon light/indoor/no windows