EXFO EXpert IPTV Test Tools (FTB-200v2) - Manual (Page 59)

59
A.2.1.4 Frame Type Detection
VQmon/HD identifies individual I, P, and B frames in the GoP and measures the packet
loss rate and loss distribution occurring in each frame type. For unencrypted video streams,
VQmon/HD performs picture header decoding to identify individual frames, GoP size, and
frame rate. For encrypted/scrambled streams, heuristic algorithms are applied in order to
detect frame boundaries and measure frame size.
As mentioned in Section 2.1.1.1, the GoP structure has impact on both the efficiency of video
encoding and the robustness of encoded video. VQmon/HD takes the different I, P, and B
frame packet loss/discard rates into account when calculating perceptual video quality
metrics.
A.2.1.5 Per‐frame Quality Analysis
VQmon/HD performs per‐frame quality calculation using the frame type, frame size,
codec type, video bandwidth, and packet loss data. The proportion of each frame type
impaired by loss/discard is reported, along with the proportion of B and P frames impaired
due to the propagation of errors from earlier reference (I or P) frames in the GoP
.
A.2.1.6 Perceptual Quality Model
VQmon/HD’s perceptual quality model calculates estimated perceptual quality (MOS)
scores using the per‐frame quality metrics and content analysis as inputs. The calculation
model considers the sensitivity of the content to quality degradation (e.g., that frame freezes
occurring during a high‐motion scene will be more visible and annoying than those occurring
during a static scene) and other subjective factors such as viewer reaction time, recency, and
temporal masking (see Section 2.1.3.2).
A.2.1.7 VQmon Markov Model (VMM)
VQmon/HD uses a four‐state Markov Model to gather and report packet loss statistics
for “burst” periods (where quality is significantly degraded) and “gap” periods (periods
between each burst interval when quality is relatively unimpaired).