Visual Media Coding and Transmission

3.3 Scalable Video Coding Techniques

3.3.1 Scalable Coding for Shape, Texture, and Depth for 3D Video

In this section, a scalable 2D-model-based video coding scheme is investigated in order to achieve the texture and depth-map coding of 3D video. The 3D video is represented by using shape, texture (color), and depth-map, as shown in Figure 3.8. Scalable coding is a popular technique for achieving robust video delivery. For example, the base layer might be heavily protected, whereas the enhancement layers would not. Error resilience can be achieved during the bursts of higher error rate by protecting the base layer at the expense of the enhancement layers. Shape-adaptive coding can facilitate the video manipulation, and keep motion and depth discontinuities, so as to improve the rendering quality of the synthesized “virtual” views.

A scalable 2D-model-based video coding scheme is presented in Figure 3.9 [1]. The coding scheme includes four parts: video segmentation, object modeling, scalable model/shape coding, and scalable texture coding. The scalable shape coding and scalable texture coding are elaborated in Subsections 3.3.1.1 and 3.3.1.2 respectively.

Figure 3.8 3D video representation by using (a) shape, (b) texture, and (c) depth-map

3.3.1.1 Scalable Shape Coding

The proposed vertex-based shape coding (both intra- and inter-) schemes are presented. In the proposed schemes, ...