Efficient coding scheme for multilevel video transmission

• US 5,253,058 A
• Filed: 04/01/1992
• Issued: 10/12/1993
• Est. Priority Date: 04/01/1992
• Status: Expired due to Term

• Current Assignee: Shingo Limited Liability Company
• Original Assignee: Bell Communications Research, Inc. (Telefonaktiebolaget LM Ericsson)
• Inventors: Gharavi, Hamid
• Primary Examiner(s): Chin, Tommy P.
• Assistant Examiner(s): LEE, RICHARD J
• Application Number:
  US07/861,627
• Time in Patent Office: 559 Days
• Field of Search: 358/133, 358/135, 358/136
• US Class Current: 375/240.12
• CPC Class Codes: H04N 19/186 the unit being a colour or a chrominance component
• CPC Class Codes: H04N 19/30 using hierarchical techniques, e.g. scalability H04N19/63 takes precedence
• CPC Class Codes: H04N 19/53 Multi-resolution motion estimation; Hierarchical motion estimation
• CPC Class Codes: H04N 19/61 in combination with predictive coding
• CPC Class Codes: H04N 19/63 using sub-band based transform, e.g. wavelets

Abstract

A highly efficient video coding scheme is disclosed which codes, for transmission, the pel values of a scanned high quality HDTV video signal in such a manner that a low-quality version of the signal as well as a high-quality version are available to different video subscribers. The coder includes a basic layer coder (110) and a contribution layer encoder (150). A subscriber to a low-quality video service receives only the basic layer signal while a high-quality video subscriber receives both the basic layer signal and the contribution layer signal, which when combined together enable the high quality video signal to be reconstructed. The basic layer coder codes the baseband of a frequency decomposed video signal using a hybrid discrete cosine transform/differential pulse code modulation coding structure, such as the CCITT recommended H.261 coder. The contribution layer coder decomposes (123) a differential high quality video signal and separately quantizes (124) and entropy codes (125) the decomposed differential subbands other than the baseband, which are then multiplexed (126) together to form the contribution layer signal. When the baseband is coded by the basic layer coder at a low bit rate, the quantization noise it introduces will deleteriously affect the quality of a high quality signal "built" upon it. A residual signal component is therefore included within the contribution layer signal. This residual layer component, which is derived in part from the basic layer coder, enables the quantization noise introduced into the basic layer to be substantially eliminated when the basic layer signal is combined with the contribution layer signal to reconstruct the high quality video signal at a receiver/decoder.

Claims

1. A signal coder for coding an input video signal into a basic layer signal and at least one contribution layer signal, a video signal having a spatial resolution lower than the spatial resolution of the input video signal being reconstructible from the basic layer signal, and a video signal having a spatial resolution higher than the lower spatial resolution and up to the resolution of the input video signal being reconstructible from the combination of the basic layer signal and the at least one contribution layer signal, the signal 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,

means for filtering the video signal to produce a baseband sequence of filtered pel values for each video frame,

interframe hybrid transform/differential pulse code modulation coding means for coding the baseband sequence of filtered pel values for each video frame to form the basic layer signal,

means for forming a differential video signal of each frame from the difference between the pel values in a current video frame and predicted pel values derived from the pel values in a previous coded video frame,

means for decomposing the differential video signal of each frame into plural subband sequences of filtered differential pel values,

means for separately quantizing and entropy coding the plural subband sequences of filtered differential pel values, and

means for combining at least some of the quantized and encoded subband sequences of filtered differential pel values to form the at least one contribution layer signal.

2. A signal coder in accordance with claim 1 wherein said means for decomposing the differential video signal of each frame is a quadrature-mirror filter.

3. A signal coder in accordance with claim 1 further comprising means for including a residual signal component within said at least one contribution layer signal, said residual signal component being formed from the combination of:

(a) a baseband signal derived from said interframe hybrid transform/differential pulse modulation means that is a reconstructed version of the baseband signal at the output of said means for filtering the video signal, (b) a baseband signal derived from the predicted pel values in a previous coded video frame from said means for forming a differential video signal, and (c) a baseband sequence of filtered differential pel values at one output of said means for decomposing the differential video signal.

4. A signal coder in accordance with claim 1 wherein said interframe hybrid transform/differential pulse code modulation coding means comprises hybrid discrete cosine transform/differential pulse code modulation coding means for producing blocks of quantized differential discrete cosine transformed baseband pel values.

5. A signal coder in accordance with claim 4 wherein said interframe hybrid transform/differential pulse code modulation coding means further comprises means for entropy coding the blocks of quantized differential discrete cosine transformed baseband pel values.

6. A signal coder in accordance with claim 4 further comprising means for forming motion estimates of frame-to-frame translations of blocks of baseband pel values in a current video frame relative to corresponding blocks of baseband pel values in a previous video frame, and means for converting the motion estimates of the blocks of baseband pel values in the current frame into motion estimates of corresponding blocks of pel values of the input video signal in the same current video frame.

7. A signal coder in accordance with claim 4 further comprising means for forming motion estimates of frame-to-frame translations of blocks of pels in the input video signal in a current video frame relative to corresponding blocks of pels in a previous video frame, and means for converting the motion estimates of the blocks of pels in the current video frame into motion estimates of corresponding blocks of pels in the baseband sequence of filtered pel values in the same current video frame.

8. A signal coder in accordance with claim 4 wherein said hybrid discrete cosine transform/differential pulse code modulation means is a CCITT recommended H.261 coder.

9. A method for coding a input video signal into a basic layer signal and at least one contribution layer signal, the input video signal consisting of sequential video frames, a video signal having a spatial resolution lower than the spatial resolution of the input video signal being reconstructible from the basic layer signal, and a video signal having a spatial resolution higher than the lower spatial resolution and up to the resolution of the input video signal being reconstructible from the combination of the basic layer signal and the at least one contribution layer signal, the method comprising the steps of:

filtering the input video signal to produce a baseband sequence of filtered pel values for each video frame,

coding the baseband sequence of filtered pel values using an interframe hybrid transform/differential pulse code modulation coder to form the basic layer signal,

forming a differential video signal of each frame from the difference between the pel values in a current video frame and predicted pel values derived from the pel values in a previous coded video frame,

decomposing the differential video signal of each frame into plural subband sequences of filtered differential pel values,

separately quantizing and encoding the plural subband sequences of filtered differential pel values, and

combining at least some of the quantized and encoded subband sequences of filtered differential pel values to form the at least one contribution layer signal.

10. The method of claim 9 wherein a quadrature-mirror filter decomposes the differential video signal into plural subband sequences of filtered differential pel values.

11. The method of claim 9 further comprising the steps of estimating frame-to-frame translations of blocks of baseband pel values, and converting the estimated frame-to-frame translations of blocks of baseband pel values into estimated frame-to-frame translations of corresponding blocks of pels in the current video frame

12. The method of claim 9 further comprising the steps of estimating frame-to-frame translations of blocks of pels in the input video signal, said estimated frame-to-frame translations being used to form the difference between the pel values in a current video frame and predicted pel values derived from pel values in a previous coded frame, and converting the estimated frame-to-frame translations of the blocks of pels in the current video frame to estimated translations of corresponding blocks of pels in the baseband sequence of filtered pel values.

13. The method of claim 9 further comprising the step of including a residual signal component within the at least one contribution layer signal, the residual signal being derived from the combination of:

(a) a baseband signal derived from the interframe hybrid transform/differential pulse code modulation coder that is a reconstructed version of the baseband signal produced at said step of filtering the input video signal, (b) a baseband signal derived from predicted pel values in a previous coded video frame produced at the step of forming a differential video signal of each frame, and (c) a baseband sequence of filtered differential pel values produced at the step of decomposing the differential video signal of each frame into plural subband sequences.

14. The method of claim 9 wherein the interframe hybrid transform/differential pulse code modulation coder is a hybrid discrete cosine transform/differential pulse code modulation coder that produces blocks of quantized differential discrete cosine transformed baseband pel values.

15. The method of claim 14 further comprising the step of entropy coding the blocks of quantized differential discrete cosine transformed baseband pel values.