User Manual

Verizon Base Unit User Guide Rev B PRELIMINARY 3/20/2014 Tech-X Flex
®
(P5)
5-32
Intro
Wi-Fi
10/100
System
IP/Video
Specs
• I-frames (or “Intra pictures”) - I-frames are coded without reference to other pictures. That is, they
contain the full dataset required to render a video frame and do not interpolate based on references
to other frames. Therefore, they may employ compression to reduce spatial redundancy, but cannot
reduce temporal redundancy. I-frames are critically important for providing references to other frames
and serve as access points in the bitstream where decoding can begin. Because other frame types
do reduce temporal redundancy based on a dependence to the I-frames, the loss of I-frames in a
video stream has the most significant impact.
• P-frames (or “Predictive pictures”) - P-frames are interspersed between I-frames and allow a
combination of spatial and temporal redundancy. They can use internal spatial coding like I-frames,
but they can also derive data through references to previous I and P-frames. Through this
referencing, a P-frame can render the picture without a full pixel-by-pixel dataset, using redundant
information presented in preceding frames.
• B-frames (or “Bi-directional predictive pictures”) - B-frames are a further extension of the P-
frame predictive methodology, except that they may reference preceding and/or following I and/or P-
frames. The use of B-frames allows the highest degree of picture quality with the most efficient
compression. When a B-frame references a frame that comes after itself, the decoder must have
received the referenced frame before the B-frame can be decoded, making the frame order different
from the actual display order. Therefore, B-frames can cause a delay in the decoding process,
because the decoder must buffer the input while reordering the frames for display. Of the three, the
loss of a B-frame generally causes the least impact to picture quality.
At the data level, a frame is divided into slices which represent horizontal sections of the frame. Each
slice is further divided into macroblocks which represent rectangular sections of the slice. This
organizational structure is the reason that digital video exhibits “rectangular” errors when data becomes
corrupted, rather than the general fuzz and/or static caused by a poor analog signal. For example:
• If macroblock data is missing or corrupted, the video typically shows rectangles of missing picture on
the screen, amidst an otherwise clear picture. Likewise, if a whole slice can’t be rendered, a larger
rectangular portion is missing.
• If whole frame data is missing or corrupted, the video may freeze on certain pictures altogether,
rendering the last known frame while waiting for new frame data.
GOP types
For any video stream, a set of frames is called a group of pictures or GOP, with the specific sequence
known as the GOP structure. A common GOP structure would include one I-frame, followed by two B-
frames, then followed by one P-frame, and so on, represented as “IBBPBBP…" The following figure
represents a simplified diagram of frame reference and interpolation, using a typical GOP structure: