Information Technology Reference
In-Depth Information
technology also introduces degradation on the image quality. Barkowsky et al.
tried to achieve a trade-off between such increased comfort and the introduced dis-
tortion [25]. However, their experimental results indicated that the distortion
caused by the depth rendering process is usually greater than the comfort provided
by the depth perception, especially for certain types of visual contents. Another
objective metric for DIBR was proposed lately in [26]. This metric consisted of
Color and Sharpness of Edge Distortion (CSED) measures. The color measure
evaluated luminance loss of the rendered image compared against the reference
image. The sharpness of edge distortion measure calculated a proportion of the
remained edge in the distorted image regarding the edge in the reference image,
whilst taking into account the depth information. To validate the performance of
the proposed metric, a subjective assessment of DIBR techniques with five differ-
ent hole-filling methods (Constant color filling, Horizontal interpolation, Region
of interest filling, Horizontal extrapolation and Horizontal and vertical interpola-
tion) was performed. The experimental results indicated the promising perform-
ance of this metric.
The rest of this chapter is organized as follows. In Section 2, a subjective quality
experiment on stereoscopic images is briefly summarized and an analysis on the re-
lationship between the perceived quality and distortion parameters is performed.
Section 3 introduces some well-known 2D image quality metrics and investigates
their capabilities in evaluating the stereoscopic image quality; an integration of dis-
parity information into objective quality metrics is proposed based on an intensive
analysis of the disparity information on stereoscopic quality evaluation. Finally,
conclusions are drawn in Section 4.
2 Subjective Stereoscopic Image Quality Assessment
The quality of a visual presentation that is meant for human consumption (the
user) can be evaluated by showing it to a human observer and asking the subject to
judge its quality on a predefined scale. This is known as subjective assessment and
is currently the most common way to evaluate the quality of image, video, and au-
dio presentations. Generally, the subjective assessment is also the most reliable
method as we are interested in evaluating quality as seen by the human eye. In this
section, we will present a subjective quality assessment on stereoscopic images,
which can be exploited for understanding perception of stereoscopic images and
providing data for designing objective quality metrics [27]. We mainly focus on
distortion types introduced by image processing but ignore the influence of a dis-
play device. The relationship between distortion parameters and the perceived
quality will be investigated, and we will also validate whether PSNR can be used
in predicting the stereoscopic image quality.
2.1 Experimental Materials and Methodology
The source stereo-pair images were collected from the Internet [28]. Ten pairs of
high resolution and high quality color images that reflect adequate diversity in
