Apparatus and method for differential sub-band coding of video signals

Claim 2

A high compression coder for coding video signals comprising: means for receiving an input signal consisting of digital PEL values of a scanned video signal, said video signal consisting of sequential video frames and each video frame consisting of a plural scan lines, said input signal having a predetermined number of PEL values per video frame, means for forming differential PEL values from digital differences between the PEL values in said input signal and digital predicted PEL values that are associated with PELs in said input signal and that are each derived from at least one previous PEL value in the input signal, means using quadrature-mirror filter means for decomposing said differential PEL values into plural narrow band sub-band sequences of filtered differential PEL values, means for discarding in each sub-band sequence of filtered differential PEL values alternate ones of the filtered values to form plural down-converted sub-band sequences of differential PEL values, the total number of down-converted differential PEL values in all of said plural down-converted sub-band sequences of filtered differential PEL values that correspond to each video frame being exactly equal to said predetermined number of PEL values per video frame, means for separately quantizing the down-converted differential PEL values in each of the plural down-converted sub-band sequences, means for entropy coding each of the quantized plural down-converted sub-band sequences of filtered differential PEL values, means for multiplexing into a single multiplexed bit stream all of said entropy coded quantized plural down-converted sub-band sequences of filtered differential PEL values, means for reconstructing said differential PEL values from said separately quantized plural down-converted sub-band sequences of filtered differential PEL values, storage means for storing reconstructed PEL values, means for combining said reconstructed differential PEL values with stored previously reconstructed PEL values to form newly reconstructed PEL values and for storing said newly reconstructed PEL values in said storage means, and means for forming said predicted PEL values from stored previously reconstructed PEL values, wherein frame-by-frame said multiplexed bit stream has substantially fewer bits than said input signal.

Claim 3

A high compression coder in accordance with claim 1 wherein said quadrature mirror filter means is a two-dimensional quadrature mirror filter.

Claim 4

A high compression coder in accordance with claim 1 wherein said means for separately quantizing comprises plural quantizers having individual quantizing schemes.

Claim 5

A high compression coder in accordance with claim 3 wherein said means for entropy coding comprises means for variable-length coding each nonzero value and means for run-length coding the location of the nonzero values in down-converted sub-band sequences of filtered differential PEL values that are associated with each scan line.

Claim 6

A high compression coder in accordance with claim 4 wherein said means for run-length coding the location of the nonzero values comprises means for coding the runs of zeros and the runs of nonzero values.

Claim 7

A high compression coder in accordance with claim 5 wherein said means for entropy coding further comprises means for producing a unique end-of-line code word following the last nonzero value in each scan line and a unique end-of-frame code word following the last nonzero value in each video frame.

Claim 8

A high compression coder in accordance with claim 3 further comprising buffer means for receiving said multiplexed bit stream and producing an output bit stream having a constant bit rate.

Claim 9

A high compression coder in accordance with claim 7 further comprising means for adjusting the quantizing schemes of said plural quantizers in accordance with the occupancy level of said buffer means.

Claim 10

A high compression coder for coding video signals comprising: means for receiving an input signal consisting of digital PEL values of a scanned video signal, said video signal consisting of sequential video frames and each video frame consisting of plural scan lines, said input signal having a predetermined number of PEL values per video frame, means for forming differential PEL values from digital differences between the PEL values in said input signal and digital predicted PEL values that are associated with PELs in said input signal and that are each derived from PEL values in a previous video frame, means using quadrature-mirror filter means for decomposing said differential PEL values into plural narrow band sub-band sequences of filtered differential PEL values, means for discarding in each sub-band sequence of filtered differential PEL values alternate ones of the filtered values to form plural down-converted sub-band sequences of differential PEL values, the total number of down-converted differential PEL values in all of said plural down-converted sub-band sequences of filtered differential PEL values that correspond to each video frame being exactly equal to said predetermined number of PEL values per video frame, means for separately quantizing the down-converted differential PEL values in each of the plural down-converted sub-band sequences, means for entropy coding each of the quantized plural down-converted sub-band sequences of filtered differential PEL values, means for multiplexing into a single multiplexed bit stream all of said entropy coded quantized plural down-converted sub-band sequences of filtered differential PEL values, means for reconstructing said differential PEL values from said separately quantized plural down-converted sub-band sequences of filtered differential PEL values, storage means for storing reconstructed PEL values, means for combining said reconstructed differential PEL values with stored previously reconstructed PEL values from the previous frame to form newly reconstructed PEL values and for storing said newly reconstructed PEL values in said storage means, and means for forming said predicted PEL values from stored reconstructed PEL values from the previous frame, wherein frame-by-frame said multiplexed bit stream has substantially fewer bits than said input signal.

Claim 11

A high compression coder for coding video signals comprising: means for receiving an input signal consisting of digital PEL values of a scanned video signal, said video signal consisting of sequential video frames and each video frame consisting of plural scan lines, said input signal having a predetermined number of PEL values per video frame, means for forming differential PEL values from digital differences between the PEL values in said input signal and digital predicted PEL values that are associated with PELs in said input signal and that are each derived from PEL values in a previous video frame, means using quadrature-mirror filter means for decomposing said differential PEL values into plural narrow band sub-band sequences of filtered differential PEL values, means for discarding in each sub-band sequence of filtered differential PEL values alternate ones of the filtered values to form plural down-converted sub-band sequences of differential PEL values, the total number of down-converted differential PEL values in all of said plural down-converted sub-band sequences of filtered differential PEL values that correspond to each video frame being exactly equal to said predetermined number of PEL values per video frame, means for separately quantizing the down-converted differential PEL values in each of the plural down-converted sub-band sequences, means for entropy coding each of the quantized plural down-converted sub-band sequences of filtered differential PEL values, means for multiplexing into a single multiplexed bit stream all of said entropy coded quantized plural down-converted sub-band sequences of filtered differential PEL values, means for reconstructing said differential PEL values from said separately quantized plural down-converted sub-band sequences of filtered differential PEL values, storage means for storing reconstructed PEL values, means for combining said reconstructed differential PEL values with stored previously reconstructed PEL values from the previous frame to form newly reconstructed PEL values and for storing said newly reconstructed PEL values in said storage means, and means for forming said predicted PEL values from stored reconstructed PEL values from the previous frame, wherein frame-by-frame said multiplexed bit stream has substantially fewer bits than said input signal.

Claim 12

A high compression coder in accordance with claim 9 wherein said quadrature mirror filter means is a two-dimensional quadrature mirror filter.

Claim 13

A high compression coder in accordance with claim 10 wherein said digital predicted PEL values are the reconstructed PEL values stored in said storage means in the corresponding spatial locations in the previous frame.

Claim 14

A high compression coder in accordance with claim 10 wherein said means for forming said predicted PEL values comprises means for estimating horizontal and vertical components of the translatory motion of corresponding PELs from one frame to the next, and said predicted PEL values are the reconstructed PEL values stored on said storage means in the corresponding spatial locations in the previous frame as shifted horizontally and vertically by the estimated motion of the PELs.

Claim 15

A high compression coder in accordance with claim 9 wherein said means for separately quantizing comprises plural quantizers having individual quantizing schemes.

Claim 16

A high compression coder in accordance with claim 4 wherein said means for entropy coding comprises means for variable-length coding each nonzero value and means for run-length coding the location of the nonzero values in down-converted sub-band sequences of filtered differential PEL values that are associated with each scan line.

Claim 17

A high compression coder in accordance with claim 14 wherein said means for run-length coding the location of the nonzero values comprises means for coding the runs of zeros and the runs of nonzero values.

Claim 18

A high compression coder in accordance with claim 15 wherein said means for entropy coding further comprises means for producing a unique end-of-line code word following the last nonzero value in each scan line and a unique end-of-frame code word following the last nonzero value in each video frame.

Claim 19

A high compression coder in accordance with claim 13 further comprising buffer means for receiving said multiplexed bit stream and producing an output bit stream having a constant bit rate.

Claim 20

A high compression coder in accordance with claim 17 further comprising means for adjusting the quantizing schemes of said plural quantizers in accordance with the occupancy level of said buffer means.

Claim 23

The method of claim 19 wherein the step of entropy coding comprises the steps of: variable-length coding in each scan line of quantized filtered values each nonzero values and run-length coding in each scan line of quantized filtered values the location of the nonzero values.

Claim 24

The method of claim 20 wherein the step of run-length coding comprises the step of run-length coding the runs of zeros and the runs of nonzero values.

Claim 25

The method of claim 21 further comprising the steps of: producing a unique end-of-line code word following the last nonzero value in each scan line, and producing a unique end-of-frame code word following the last nonzero value in the video frame.

Claim 26

The method for compressing an input bit stream containing the digital PEL values of a scanned video signal of claim 19 further comprising the step of: storing the multiplexed bit stream in a buffer which produces an output at a constant bit rate.

Claim 27

The method of claim 23 further comprising the step of: adjusting the quantizing of the plural narrow band digital signals in accordance with the occupancy level of the buffer.

Claim 28

A high compression video coder comprising means for receiving an input signal consisting of PEL values of a scanned video signal, said video signal consisting of sequential video frames and each video frame consisting of plural scan lines, said input signal having a predetermined number of PEL values per video frame, means for forming differential PEL values from differences between the PEL values in the input signal and predicted PEL values that are associated with PELs in the input signal and that are each derived from at least one previous PEL value in the input signal, means using quadrature-mirror filter means for decomposing said differential PEL values into plural narrow band sub-band sequences of filtered differential PEL values, means for discarding in each sub-band sequence of filtered differential PEL values alternate ones of the filtered values to form plural down-converted sub-band sequences of differential PEL values, the total number of down-converted differential PEL values in all of said plural down-converted sub-band sequences that correspond to each video frame being exactly equal to said predetermined number of PEL values per video frame, means for separately quantizing the down-converted differential PEL values in each of the plural down-converted sub-band sequences, means for entropy coding each of the quantized plural down-converted sub-band sequences of filtered differential PEL values, and means for multiplexing into a single multiplexed bit stream all of said entropy coded quantized plural down-converted sub-band sequences of filtered differential PEL values.

Claim 29

A high compression video coder comprising means for receiving an input signal consisting of PEL values of a scanned video signal, said video signal consisting of sequential video frames and each video frame consisting of plural scan lines, said input signal having a predetermined number of PEL values per video frame, means for forming differential PEL values from differences between the PEL values in the input signal and predicted PEL values that are associated with PELs in the input signal and that are each derived from at least one previous PEL value in the input signal, means using quadrature-mirror filter means for decomposing said differential PEL values into plural narrow band sub-band sequences of filtered differential PEL values, means for discarding in each sub-band sequence of filtered differential PEL values alternate ones of the filtered values to form plural down-converted sub-band sequences of differential PEL values, the total number of down-converted differential PEL values in all of said plural down-converted sub-band sequences that correspond to each video frame being exactly equal to said predetermined number of PEL values per video frame, means for separately quantizing the down-converted differential PEL values in each of the plural down-converted sub-band sequences, means for entropy coding each of the quantized plural down-converted sub-band sequences of filtered differential PEL values, and means for multiplexing into a single multiplexed bit stream all of said entropy coded quantized plural down-converted sub-band sequences of filtered differential PEL values.

Claim 30

A high compression video coder in accordance with claim 25 wherein said quadrature mirror filter means is a two-dimensional quadrature mirror filter.

Claim 31

A high compression video coder in accordance with claim 26 wherein said means for separately quantizing comprises plural quantizers having individual quantizing schemes.

Claim 32

A high compression coder in accordance with claim 27 wherein said means for entropy coding comprises means for variable-length coding each nonzero value and means for run-length coding the location of the nonzero values in down-converted sub-band sequences of filtered differential PEL values that are associated with each scan line.

Claim 33

A high compression coder in accordance with claim 28 wherein said means for run-length coding the location of the nonzero values comprises means for coding the runs of zeros and the runs of nonzero values.

Claim 34

A high compression coder in accordance with claim 29 wherein said means for entropy coding further comprises means for producing a unique end-of-line code word following the last nonzero value in each scan line and a unique end-of-frame code word following the last nonzero value in each video frame.

Claim 35

A high compression video coder in accordance with claim 27 further comprising buffer means for receiving said multiplexed bit stream and producing an output bit stream at a constant bit rate.

Claim 36

A high compression video coder in accordance with claim 31 further comprising means for adjusting the quantizing schemes of said plural quantizers in accordance with the occupancy level of said buffer means.