Autofocus on Depth of Interest for 3D Image Coding

Figure 26

Visual quality comparison of synthesized view for (a) UndoDancer view 3 frame 250 using (b) uncompressed depth and texture; depth coded at 0.014 bpp and texture coded at 0.2 bpp (c) by classic LAR; (d) by the proposed DoI refinement of depth scheme and SNR quality enhancement of texture with a full-resolution DoI-Mask; (e) by the proposed DoI scheme with the block resolution DoI-Mask.

(a)

(b)

(c)

(d)

(e)