Instruction Manual

54
IPTV applications generally support video encoded with SD resolutions of 720x486 (30 Hz) or
720x576 (25 Hz), or HD resolutions of 1280x720 (720p) or 1920x1080 (1080i) at an average
bitrate of 3 Mbits/sec or higher.
Typically, the number of bits spent on coding a P-frame is about 20% of the number required
for an I-frame, whereas a B-frame takes about 5% of the bitrate consumption of the I-frame.
This implies that group of pictures (GOP) structure has a big impact on quality of encoded
video at a given bitrate under no loss condition, as well as under loss conditions as described
below. VQmon/HD automatically adjusts the video stream effective bitrate based on the GOP
structure of the encoded video stream.
A.1.3 Impact of Subjective Factors on Perceptual Quality
The degree to which viewers find video impairments annoying—or notice them at all—
depends in part on the severity and duration of the impairment events, but also on certain
inherent characteristics of human perception. The same type of impairment may be extremely
irritating or barely noticed, depending on factors such as the scene content when the error
takes place, and whether it occurs alone or simultaneously with other impairments.
A.1.3.1 Video Content
The visibility of video problems depends partially on the scene content; for example, frame
freezes tend to be much more noticeable in sequences containing high levels of motion than
in relatively static scenes, such as footage of a television news anchor. VQmon/HD performs
content and scene analysis, detecting levels of detail, motion, and panning, and can detect
and provide alerts for content problems such as noise/snow and frame freezes. Scene
analysis data is leveraged to increase the accuracy of VQmon/HD’s estimated perceptual
quality scores.
A.1.3.2 Temporal Phenomena: Reaction Time, Masking and Recency Effects
Perceptual quality is affected somewhat by a natural delay in human reaction time when
impairments occur, or when quality improves after a period of degradation; i.e., the viewer’s
reaction to either event is not immediate. As with audio, when two or more impairments
occur simultaneously (or in rapid succession), there can be a “masking” effect that affects the
way viewers perceive the severity of quality degradation. In addition, a “recency”
phenomenon exists, wherein viewers tend to perceive impairments as more severe when they
occurred recently, but are willing to “forgive” them to some extent as time passes.