GSO Algorithm

Kendo (.pdf)

Note: Click on images for a larger view.

Plots PSNR vs. bitrate for the View 2 color image synthesized using the color and depth image of two other views (View 1 and View 3) for dataset Kendo. The depth images depth_1 (from View 1) and depth_3 (from View 3) are compressed in turn with the lossy methods "GSOm+CCV", "GSOs+CCLV", "GSOm+CERV", "GSOm+PWC" and JPEG 2000, and the compressed versions are fed to the View Synthesis Reference Software (VSRS), which will create for each lossy pair (depth_1,depth_3) at a certain bitrate, a synthesized color image for View 2. Using the lossless pair (depth_1,depth_3) one obtains a synthesized image "GroundTruthSynthView", which is taken as a reference in the computation of the PSNR. The variants of GSO algorithm are showing better performance than the JPEG2000, and the shape of the curves are similar with the simple PSNR vs. bitrate curve of the depth images shown in Figures 1(c)-(i) of the letter.
Last update: August 16, 2013
Designed by: Ionut Schiopu