Quality Assessment - 3DTV: Processing and Transmission of 3D Video Signals - page 164

3D video. Objective 3D quality assessment is still far from being mature.
Objective video quality assessment (for both 3D and 2D) is generally classified
as follows:

• Full-reference objective video quality assessment (FR);

• No-reference objective video quality assessment (NR);

• Reduced-reference objective video quality assessment (RR).

In FR video quality assessment it is assumed that the quality assessment
technique has full access to both processed and original versions of the video
sequence. The typical functions of a FR assessment model are described in
Figure 6.10(a). In some cases, such as video quality evaluation on the receiver
side, FR methods may not be applicable as the reference video sequence is
not available. On the other hand, NR video quality assessment uses only the
information from the processed video to produce an objective quality score.
These models do not have access to the reference video. However, they use
prior information about the distortion of the video sequence. For instance,

Reference video

Quality rating

FR objective quality
assessment model

Proces sed video

(a)

Blockiness

Distortions
measurement
(blockiness,
blur, etc…)

Blur

Distortion
pooling

Quality rating

Pro cessed video

Etc.

NR objective quality assessment model

(b)

Referenc e video

Feature
extraction

Quality rating

RR objective quality
assessment model

Processed video

(c)

Figure 6.10 Objective quality assessment techniques. (a) FR objective quality
assessment model (b) NR objective quality assessment model (c) RR objective
quality assessment model