P1: IML/FFX P2: IML/FFX QC: IML/FFX T1: IML
Video ̇Compression ̇OLE WL040/Bidgolio-Vol I WL040-Sample.cls September 14, 2003 18:10 Char Count= 0
542 VIDEOCOMPRESSIONvision processing. Figure 2b shows the Generalized Bal-
anced Ternary sampling and addressing scheme. Seven
hexagons, labeled from 0 to 6, form an aggregate that can
be used to tile the plane. The address “23” refers to the
hexagon in position 3 of the aggregate in position 2. If we
define the address 7 to refer to an aggregate of hexagons,
the address “077” indicates a pattern formed from the ag-
gregation of aggregates of hexagons.
See Snyder, Qi, and Sander (1999) and the exploration
of tesseral addressing schemes advocated by Bell, Diaz,
Holroyd, and Jackson (1983), and Jackson, Bell, Stevens,
Freedman, and Dickman (1986).DIGITAL VIDEO SIGNAL
COMPRESSION
Overview and Tradeoff Dimensions
The video compression process typically trades commu-
nications or storage bandwidth for processor cycles and
picture quality determined by the intended application of
the digital video sequence. In addition to considerations of
image compression applied to each frame independently
of the others (intraframe coding), scenes showing mod-
erate motion may be effectively compressed by consider-
ing what has changed between frames (interframe coding
and conditional replenishment). A motion picture will of-
ten have video sequences synchronized with audio tracks
and ancillary information such as system parameters. A
video sequence may be coded at multiple bit rates or in
multiple versions.Digital Video Quality Assessment
Measuring the quality of a video sequence reconstructed
from compressed imagery is an application-dependent
task. While there may be well-defined criteria for accept-
ability of the compressed video sequence intended for
machine use in a scientific or medical discipline, consid-
erable research activity is still needed to close the gap
between objective and subjective assessments of quality
for video sequences intended for human viewing.Human Visual Response
Assessment of video quality by human observers, invited
to view scenes encoded with varying parameters takes
the human visual response (HVR) explicitly into account.
Jain (1989) and the HyperPhysics Web site (Nave, 2001)
introduced the following key elements of the HVR.Eye movements focus scenes onto the retina, which con-
tains rod and cone photoreceptors. The cones are clus-
tered about the center of the retina (fovea centralis)
and are sensitive to red, blue, and green color stimu-
lus (photopic vision). Although the blue-sensitive cones
number much less than the red- or green-sensitive
cones, their sensitivity is far greater. The rods are highly
sensitive to peripheral motion at low light levels and
relatively insensitive to color (scotopic vision).
Humans perceive luminance in terms of relative contrast
(Weber’s Law).The Mach Band effect demonstrates that perceived bright-
ness is not a monotonic function of luminance (lateral
inhibition).
MacAdam ellipses demonstrate regions ofjust noticeable
differencein color in the CIE chromaticity diagram.
Bloch’s Law states that “light flashes of different durations
but equal energy are indistinguishable below a critical
duration.” This critical duration depends on how well
the eye is adapted to the dark and is nominally about
30 ms.
When a light flashes at a rate above the critical fusion fre-
quency, the flashes are indistinguishable from a steady
light of average intensity.
The eye is more sensitive to flickering of high spatial fre-
quencies than low spatial frequencies.
Human attention is attracted by faces in video scenes but
may be distracted by peripheral object motion.Subjective Evaluation
Subjective evaluation requires a group of human
observers—preferably not expert in image quality
assessment—to view and rate video quality in terms of a
scale of impairments, which range from “not noticeable”
to “extremely objectionable.” ITU-R BT.500–10, Method-
ology for the Subjective Assessment of the Quality of Tele-
vision Pictures, recommends a specific system prescribing
viewing conditions, range of luminance presented to the
viewer panel, number and experience of viewers, moni-
tor contrast, selection of test materials, and process for
evaluation of test results. The Double Stimulus Impair-
ment Scale and the Double Stimulus Continuous Qual-
ity Scale are particularly noteworthy. Subjective tests are
costly and not highly reproducible. ANSI T1.801.01–1995
provides a set of test scenes in digital format while ANSI
T1.801.01–1996 provides a dictionary of commonly used
video quality impairment terms.Objective Evaluation
Objective evaluation techniques range from simple test
metrics that do not take the HVR into account (such as the
peak signal-to-noise ratio often quoted by researchers) to
the vision system model metric developed by the Sarnoff
Corporation (Sarnoff Corporation, 2001), which relies
on comparison of maps of just noticeable differences
between original and compressed video sequences, one
frame at a time. Annex A of ANSI T1.803.03–1996 lists a
set of objective test criteria that may be used to measure
video quality in one-way video systems, applying objec-
tive tests closely related to known features of the HVR.
Webster et al. (1993) presented a scheme for combining
subjective and objective assessments on test scenes based
on objectively generated impairments.Rate Distortion Relationships
The Shannon (1948) rate distortion bound refers to the
minimum average bit rate required to encode a data
source for a given average distortion level. If the data
can be perfectly reconstructed, the bit rate at zero dis-
tortion is equal to the source entropy, a measure of the
information contained in the source. In practice, many