Information Technology Reference
In-Depth Information
60
50
40
30
20
10
0
0
10
20
30
40
50
60
Optimum predictive quality values
Fig. 9 Scatter plot of DMOS versus optimum predictive quality values
In summary, our experiments indicate that 2D IQMs can not be adopted in
evaluating the stereoscopic image quality directly, and the disparity information
has a significant impact on the perceived quality. The future work is needed to ex-
plore the relationship between the original image, the disparity information, and
the quality assessment in depth.
4 Conclusions
In this chapter, we have investigated related issues in visual quality assessment for
3D presentations, especially the stereoscopic image. Typical distortion types on
3D presentations introduced from content capture, coding schemes, to transmis-
sion through communication channels, and in displaying the 3D presentation on an
auto-stereoscopic display, were reviewed. We mainly focused on an analysis of
the quality degradation caused by coding errors. To study the relationship between
the perceived quality and distortion conditions for the stereoscopic images, a sub-
jective quality assessment was conducted. Four typical distortion types: Gaussian
blur, JPEG compression, JPEG2000 compression, and white noise, were intro-
duced to some popular stereoscopic images, and the subjective quality evaluation
was conducted in a controlled laboratory. We performed an intensive analysis on
the relationship between the perceived quality and distortion conditions on the
stereoscopic images. It was found that the perceived quality is dependent strongly
on the distortion type and image content. The performance of PSNR in predicting
the stereoscopic image quality was evaluated with respect to the subjective results.
However, it was found that PSNR is not an appropriate metric for the stereoscopic
image quality assessment. Therefore, we investigated the capabilities of some
Search WWH ::




Custom Search