Quality Assessment - 3DTV: Processing and Transmission of 3D Video Signals - page 165

they know the type of encoding scheme that was used, in order to look for
codec-specific distortion, such as blockiness, blur, etc. Therefore, typically
NR quality models are designed for one or a set of specific distortion types
and are unlikely to generalize for assessment of other types of distortion. The
NR quality assessment model is described in Figure 6.10(b).
RR video quality assessment provides a solution that lies between FR and
NR assessment. In RR quality assessment, certain features that are extracted
from the reference video are available for the quality assessment model to
evaluate the quality of the processed video. In a practical context, these
important features must be coded and transmitted with the compressed
video produced by the encoder to the RR quality assessment model. In
general, the extracted features are coded using a much lower data rate than
the compressed video and are transmitted through an error-free channel or
in the same channel but with better protection (e.g. through error control
coding) [14]. RR quality assessment is illustrated in Figure 6.10(c).
The goal of objective 3D video quality assessment research is to design
effective models to automatically predict the quality attributes of 3D video.
This will enable service providers, developers and the standards organi-
zations to rely on meaningful quality evaluation methodologies without
resorting to full subjective tests.
6.3.1 SubjectiveandObjectiveQualityMeasurements
Efficient compression techniques are vital for bandwidth-limited communi-
cation channels. Regardless of the coding methods used, rendered 3D video
quality is affected by the artefacts introduced to the colour texture and depth
sequences during compression. 3D video shows some new classes of quality
features in addition to all of the quality features of 2D video. As discussed
in Section 6.2, the quality of experience of 3D video is multi-dimensional
in nature. It can be described as a combination of several attributes such
as image quality, depth quality and visual comfort. The impairments (e.g.
blockiness, noise) introduced by image/video compression, produce spe-
cific 3D quality artefacts (e.g. cross-talk, discomfort) in the reconstructed 3D
video. This chapter investigates and describes an approach for modelling
two dominant perceptual quality attributes, i.e. image and depth quality,
of 3D video from the user perspective. Several subjective experiments are
conducted to investigate and model the above perceptual quality attributes
of 3D video. The common test conditions used for all tests described in this
chapter are given below.
The evaluation of video quality can be divided into two classes: subjective
and objective methods. Intuitively one can say that the best judgement of
quality is the human being. This is why subjective methods are said to
be the most precise measures of perceptual quality, and to date subjective