Efficiency FGST framework employing higher quality reference frames

A method of coding a video signal, comprising the steps of:

• coding the video signal into a data-stream of base layer frames;
• computing residual image frames in a transform domain from the base layer frames;
• constructing extended base layer reference frames from the base layer frames and at least portions of the residual image frames;
• motion-compensating the extended base layer reference frames to produce motion-compensated extended base layer reference frames;
• predicting motion-compensated residual image frames from the motion-compensated extended base layer reference frames and the video signal;
• and
• fine granular scalable coding the motion-compensated residual image frames into a data-stream of temporal frames.

An apparatus for coding video, comprising:

• means for coding the video into a data-stream of base layer frames;
• means for computing residual image frames in a transform domain from the base layer frames;
• means for constructing extended base layer reference frames from the base layer frames and at least portions of the residual image frames;
• means for motion-compensating the extended base layer reference frames to produce motion-compensated extended base layer reference frames;
• means for predicting motion-compensated residual image frames from the motion-compensated extended base layer reference frames and the video;
• and
• means for fine granular scalable coding the motion-compensated residual image frames into a data-stream of temporal frames.

A memory medium for coding video, comprising:

• code for coding the video into a data-stream of base layer frames;
• code for computing residual image frames in a transform domain from the base layer frames;
• code for constructing extended base layer reference frames from the base layer frames and at least portions of the residual image frames;
• code for motion-compensating the extended base layer reference frames to produce motion-compensated extended base layer reference frames;
• code for predicting motion-compensated residual image frames from the motion-compensated extended base layer reference frames and the video;
• and
• code for fine granular scalable coding the motion-compensated residual image frames into a data-stream of temporal frames.

A method of decoding a base layer data-stream and an enhancement layer date-stream, which collectively represent a video, comprising the steps of:

• decoding the base layer data-stream to construct base layer frames;
• fine granular scalable decoding an SNR quality frame portion of the enhancement layer data-stream to construct SNR quality frames;
• constructing extended base layer reference frames from the base layer frames and at least portions of the SNR quality frames;
• motion-compensating the extended base layer reference frames to produce motion-compensated extended base layer reference frames;
• fine granular scalable decoding a temporal frame portion of the enhancement layer data-stream to construct motion-compensated residual frames;
• combining the motion-compensated extended base layer reference frames with the motion-compensated residual frames to construct temporal frames.

An apparatus for decoding a base layer data-stream and an enhancement layer data-stream, which collectively represent a video, comprising:

• means for decoding the base layer data-stream to construct base layer frames;
• means for fine granular scalable decoding an SNR quality frame portion of the enhancement layer data-stream to construct SNR quality frames;
• means for constructing extended base layer reference frames from the base layer frames and at least portions of the SNR quality frames;
• means for motion-compensating the extended base layer reference frames to produce motion-compensated extended base layer reference frames;
• means for fine granular scalable decoding a temporal frame portion of the enhancement layer data-stream to construct motion-compensated residual frames;
• means for combining the motion-compensated extended base layer reference frames with the motion-compensated residual frames to construct temporal frames.

A memory medium for decoding a base layer data-stream and an enhancement layer data-stream, which collectively represent a video, comprising:

• code for decoding the base layer data-stream to construct base layer frames;
• code for fine granular scalable decoding an SNR quality frame portion of the enhancement layer data-stream to construct SNR quality frames;
• code for constructing extended base layer reference frames from the base layer frames and at least portions of the SNR quality frames;
• code for motion-compensating the extended base layer reference frames to produce motion-compensated extended base layer reference frames;
• code for fine granular scalable decoding a temporal frame portion of the enhancement layer data-stream to construct motion-compensated residual frames;
• code for combining the motion-compensated extended base layer reference frames with the motion-compensated residual frames to construct temporal frames.