Methods and devices for self-correcting video compression

DETAILED DESCRIPTION The inventive METHOD AND DEVICES FOR SELF-CORRECTING VIDEO COMPRESSION has as its objects: (a) to more efficiently encode sequences of images, so that less data is needed to transmit or store higher quality images; (b) to eliminate encoding of the images on the basis of the differences between frames in the input, such encoding results in error build up, and therefore eliminating this encoding eliminates the need to intersperse non- differentially encoded frames into sequences of differentially encoded frames; (c) to serve as the means of encoding frames whether they are consecutive or are the beginning and end frames of a sequence with consistent motion; (d) to permit the user to determine the type and amount of information to be discarded; (e) to eliminate the need to discard some of the data so that the remaining data will efficiently fit into computer memories; (f) to eliminate the truncating errors; and (g) to provide a system that can identically reproduce all frames and can readily trade off the length of the encoded message for the degree of differences allowed.
To achieve these objects, this invention makes two significant improvements.
One, it encodes the difference between an input and the effect of the previous input, rather than the difference between two inputs.
Thus, in effect, this invention models a feedback system, rather than an open loop system, and gains the well known advantages of feedback systems.
And two, it does not transform the pixels, but processes them in their natural state to prevent the errors related to the transformations.
This invention primarily has the encoder generate information essentially identical to the information generated by the decoder ("decoder information") at each stage.
The encoder computes an output signal that one or more decoders will receive reflecting the difference between the encoded instant frame and the decoder information just prior to operating on the instant frame.
Therefore, the version of the instant frame in the encoder can be fully reproduced by the decoder