Information Technology Reference
In-Depth Information
image contents were chosen. Figure 1 shows parts of the right eye images in the
data set. For each pair of the source images, the right eye image was distorted with
four different distortion types that may occur in real-world applications while the
left eye image was kept undistorted. The distortion types included:
•
Gaussian blurring: The R, G, and B color components were filtered using a cir-
cular-symmetric 2-D Gaussian kernel of standard deviation σB pixels. Three
color components of an image were blurred using the same kernel, and σB
values ranged from 0.2 to 100 pixels.
•
JPEG compression: The distorted images were generated by compressing the
reference images (full color) with JPEG at different bit rates ranging from 0.15
bits per pixel (bpp) to 3.34 bpp. The compression was implemented by a
MATLAB's toolbox function (imwrite.m).
•
JPEG2000 compression: The distorted images were generated by compressing
the reference images (full color) with JPEG2000 at different bit rates ranging
from 0.003 bpp to 2 bpp. Kakadu version 2.2 [29] was used to generate the
JPEG2000 compressed images.
•
White noise: White Gaussian noise with standard deviation σN was added to
RGB color components of the images after scaling these three color compo-
nents between 0 and 1. The used values of σN were between 0.012 and 2.0. The
distorted components were clipped between 0 and 1, and then re-scaled to a
range of [0-255].
Fig. 1
Examples of stereoscopic images (top left:
Art
; top right:
Bowling
; bottom left:
Dwarves
; bottom right:
Moebius
)
