Every paper I read about statistical modelling of video data frame size, packet level, investigates each frame type separately. That leads to assigning different statistical models for each frame type, I, P & B of the same data sample file. However, video encoding, specifically H.264, relates the three frames in a specific location pattern. Isn't that a reason for sharing sufficient statistical parameters that entitles a common statistical distribution?
Thank you Aparna for your reply.
The link provided with point 1 seems to be explaining the benefit of using inter and intra frame prediction in the encoder level. I am referring to transmitted video data that is received as packets to the end user via internet protocol.
Also, will you explain more about point 3. How does having 3 frame types provide better prediction, are you referring to prediction of frame size? and how?
After doing more research on the subject. I believe that the inherited relation of the three frame' types encoding enables us to relate their distribution. However, we need to start with finding a suitable distribution for the I frame, since it is the intra coded frame type, and use that to model the distribution of the P & B frames.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling