Re: Painting Initial Frame

Bob (
Tue, 7 Nov 1995 08:00:08 -0600

> How could the reflector do this? It does not have a copy of the frame.
>It is just passing on packets.
It may be passing on packets "now". There is no reason that the reflector
could not maintain the current "frame" for each sender, just the same as the
senders and receivers do.
>> Since the relevant data will tend to be centered in the frame, methodically
>> paint the frame rectangularly from the center out. On the first pass, omit
>> sending initial squares which "closely" match the intial solid frame.
>> Consider halving or quartering the information in every 1st square sent, and
>> dither back at the receiver. They will be cleaned up later when the square
>> is sent again. Getting the initial image in some kind of "correct"
>> decipherable format is the primary concern. The brain will "clean it up",
>> in the sense that the image will be better defined and integratable earlier
>> than it is now. Keep track of where the face is (pattern recognition).
>> When behind in data transmission, put a priority on the facial squares.
> How can the reflector do this when (1) computer recognition of the face
>is still an unsolved problem, and (2) the reflector does not even have a
>copy of the frame locally.
1) Computer recognition of an INDIVIDUAL face is unsolved (not really).
But, that is not necessary here. Recognition of the area of an image in
which a face (any face) is located is a SOLVED problem.

2) The reflector COULD have the frame locally.

What I am suggesting is an intelligent reflector, not a packet-passer...