Method and apparatus for predictive coding of moving images while adaptively changing the method in accordance with them

• US 5,418,569 A
• Filed: 06/09/1993
• Issued: 05/23/1995
• Est. Priority Date: 06/10/1992
• Status: Expired due to Fees

• Current Assignee: Victor Company of Japan Limited (JVC Kenwood Corporation)
• Original Assignee: Victor Company of Japan Limited (JVC Kenwood Corporation)
• Inventors: Ando, Ichiro
• Primary Examiner(s): Chin, Tommy P.
• Assistant Examiner(s): LEE, RICHARD J
• Application Number:
  US08/073,906
• Time in Patent Office: 713 Days
• Field of Search: 358/133, 358/136
• US Class Current: 375/240.04
• CPC Class Codes: H04N 19/124 Quantisation
• CPC Class Codes: H04N 19/132 Sampling, masking or truncation of coding units, e.g. adaptive resampling, frame skipping, frame interpolation or high-frequency transform coefficient masking
• CPC Class Codes: H04N 19/136 Incoming video signal characteristics or properties
• CPC Class Codes: H04N 19/146 Data rate or code amount at the encoder output
• CPC Class Codes: H04N 19/149 by estimating the code amount by means of a model, e.g. mathematical model or statistical model
• CPC Class Codes: H04N 19/15 by monitoring actual compressed data size at the memory before deciding storage at the transmission buffer
• CPC Class Codes: H04N 19/177 the unit being a group of pictures [GOP]
• CPC Class Codes: H04N 19/527 Global motion vector estimation
• CPC Class Codes: H04N 19/587 involving temporal sub-sampling or interpolation, e.g. decimation or subsequent interpolation of pictures in a video sequence
• CPC Class Codes: H04N 19/61 in combination with predictive coding
• CPC Class Codes: H04N 19/86 involving reduction of coding artifacts, e.g. of blockiness

Abstract

A method for predictive coding of moving images consists of the steps of calculating preliminary prediction errors of an original group of frames of moving images, accumulating the preliminary prediction errors to obtain an accumulated value, omitting one or more frames from the original group of frames to produce a decimated group of frames of moving images in cases where the accumulated value is larger than a modified order threshold value Th₂, weighting quantization weighting factors in cases where the accumulated value is larger than a decimated order threshold value TH₃, calculating prediction errors of the decimated group of frames, quantizing the prediction errors with the quantization weighting factors, encoding the prediction errors to produce pieces of output data of which each are composed of a plurality of codes. The amount of the codes of the output data is lower than an upper limit, and the decimated group of frames of moving images is distinctly reproduced by decoding the codes of the output data.

Claims

1. A method for encoding a plurality of prediction errors of a group of frames of moving images, comprising the steps of:

calculating preliminary prediction errors of an original group of frames arranged in original order, a frame of moving image existing in each of the frames, and each of the preliminary prediction errors being defined as a difference between a frame of moving image actually produced and a frame of moving image predicted;

accumulating the preliminary prediction errors of the original group of frames to obtain an accumulated value;

omitting one or more frames from the original group of frames to produce a decimated group of frames arranged in decimated order in cases where the accumulated value is larger than a modified order threshold value Th₂ ;

calculating prediction errors of the decimated group of frames, each of the prediction errors of the decimated group of frames being defined as a difference between a frame of moving image actually produced and a frame of moving image predicted, and the sum of the prediction errors of the decimated group of frames being smaller than the modified order threshold value Th₂ ; and

encoding the prediction errors of the decimated group of frames to produce pieces of output data of which each are composed of a plurality of codes, the amount of the codes of the output data being lower than an upper limit, and the decimated group of frames of moving images being reproduced by decoding the codes of the output data.

2. A method according to claim 1, additionally including:

weighting quantization weighting factors in cases where the accumulated value is larger than a decimated order threshold value Th₃, the decimated order threshold value Th₃ being larger than the modified order threshold value Th₂ ; and

quantizing the prediction errors of the decimated group of frames with the quantization weighting factors, prior to the step of encoding the prediction errors, the amount of codes obtained by encoding the prediction errors quantized with the quantization weighting factors being smaller than the amount of codes obtained by encoding the prediction errors quantized without the quantization weighting factors.

3. A method according to claim 1 in which the step of omitting one or more frames includes:

replacing the frames omitted with a frame preceding the frames omitted to equalize the number of frames in the decimated group to the number of frames in the original group.

4. A method according to claim 1 in which the step of omitting one or more frames includes:

replacing the frames omitted with a frame succeeding the frames omitted to equalize the number of frames in the decimated group to the number of frames in the original group.

5. A method according to claim 2 in which the step of weighting quantization weighting factors includes:

heavily weighting the quantization weighting factors, relating to higher frequency elements to which a viewer is insensitive, on priority basis; and

slightly weighting the quantization weighting factors relating to lower frequency elements to which a viewer is sensitive.

6. A method for encoding a plurality of prediction errors of a group of frames, comprising the steps of:

preparing an original group of frames composed of one or more I-frames, one or more P-frames, and one or more B-frames, a frame of moving image existing in each of the I-, P- and B-frames, an I-frame of moving image being not predicted with any frame, a P-frame of moving image being predicted with an I-frame or another P-frame of moving image preceding the P-frame of moving image, and a B-frame of moving image being predicted with both an I-frame or a P-frame of moving image preceding the B-frame of moving image and an I-frame or a P-frame of moving image succeeding the B-frame of moving image;

preparing a modified group of frames composed of one or more I-frames, one or more P-frames, and one or more B-frames, the number of B-frames being increased while decreasing the number of P-frames as compared with the original group of frames, and the number of the modified group of frames being equal to the number of the original group of frames;

preparing a slightly decimated group of frames which is produced by slightly omitting one or more frames from the original group of frames and replacing each of the frames omitted with a frame preceding or succeeding the frames omitted;

preparing a heavily decimated group of frames which is produced by heavily omitting a plurality of frames from the modified group of frames and replacing the frames omitted with a frame preceding or succeeding the frames omitted;

calculating preliminary prediction errors of the original group of frames arranged in original order, each of the preliminary prediction errors being defined as a difference between a frame of moving image actually produced and a frame of moving image predicted;

accumulating the preliminary prediction errors of the original group of frames to obtain an accumulated value;

calculating first prediction errors of the original group of frames in cases where the accumulated value is smaller than an original order threshold value The, each of the first prediction errors of the original group of frames being defined as a difference between a frame of moving image actually produced and a frame of moving image predicted;

calculating second prediction errors of the modified group of frames in cases where the accumulated value ranges between the original order threshold value Th_m and a modified order threshold value Th₂ larger than the original order threshold value Th₁, each of the second prediction errors of the modified group of frames being defined as a difference between a frame of moving image actually produced and a frame of moving image predicted;

calculating third prediction errors of the slightly decimated group of frames in cases where the accumulated value ranges between the modified order threshold value Th₂ and a decimated order threshold value Th₃ larger than the modified order threshold value Th₂, each of the third prediction errors of the slightly decimated group of frames being defined as a difference between a frame of moving image actually produced and a frame of moving image predicted;

calculating fourth prediction errors of the heavily decimated group of frames in cases where the accumulated value is larger than the decimated order threshold value Th₃, each of the fourth prediction errors of the heavily decimated group of frames being defined as a difference between a frame of moving image actually produced and a frame of moving image predicted; and

encoding the first prediction errors, the second prediction errors, the third prediction errors or the fourth prediction errors to produce pieces of output data of which each are composed of a plurality of codes, the amount of the codes of the output data being lower than an upper limit, and one group of frames of moving images selected from among the original group of frames of moving images, the modified group of frames of moving images, the slightly decimated group of frames of moving images, and the heavily decimated group of frames of moving images being reproduced by decoding the codes of the output data.

7. A method according to claim 6, additionally including:

weighting quantization weighting factors in cases where the accumulated value is larger than the decimated order threshold value Th₃ ; and

quantizing the fourth prediction errors of the heavily decimated group of frames with the quantization weighting factors, prior to the step of encoding the prediction errors, the amount of codes obtained by encoding the fourth prediction errors of the heavily decimated group of frames quantized with the quantization weighting factors being smaller than the amount of codes obtained by encoding the fourth prediction errors of the heavily decimated group of frames quantized without the quantization weighting factors.

8. A method according to claim 7 in which the step of weighting quantization weighting factors includes:

heavily weighting the quantization weighting factors, relating to higher frequency elements to which a viewer is insensitive, on priority basis; and

slightly weighting the quantization weighting factors relating to lower frequency elements to which a viewer is sensitive.

9. A method for encoding successive frames of moving images, in which the successive frames are divided into groups consisting of a predetermined number of frames, the method comprising the step of:

calculating inter-frame prediction errors which each are derived from a difference between a current frame belonging to a current group of an original order and a reference frame adjacent to the current frame;

calculating a sum of the inter-frame prediction errors before the frames belonging to the current group are encoded;

comparing the sum of the inter-frame prediction errors with a modified order threshold level Th₂ ;

reconstituting a new current group of frames by omitting at least one frame from the current group in cases where the sum exceeds the modified order threshold level Th₂ ; and

respectively encoding frames belonging to the new current group.

10. A method according to claim 9, in which the step of respectively encoding frames includes:

calculating inter-frame prediction errors which each are derived from a difference between a new current frame belonging to the new current group and a reference frame adjacent to the new current frame;

orthogonally transforming the inter-frame prediction errors calculated into transformed coefficients using discrete cosine transform;

quantizing the transformed coefficients into digital data by first weighting factors; and

converting the digital data into predetermined codes.

11. A method according to claim 9, in which the step of respectively encoding frames includes:

comparing the sum with a decimated order threshold level Th₃ which is larger than the modified order threshold level Th₂ ;

calculating inter-frame prediction errors which each are derived from a differences between a new current frame belonging to the new current group and a reference frame adjacent to the new current frame;

orthogonally transforming the inter-frame prediction errors calculated into transformed coefficients using discrete cosine transform;

quantizing the transformed coefficients into digital data by second weighting factors in cases where the sum exceeds the decimated order threshold level Th₃ ; and

converting the digital data into predetermined codes.

12. A moving image encoding apparatus, comprising:

prediction error accumulating means for accumulating preliminary prediction errors of an original group of frames of an original or to obtain an accumulated value, a frame of moving image existing in each of the frames, and each of the preliminary prediction errors being defined as a difference between a frame of moving image actually produced and a frame of moving image predicted;

frames omitting information generating means for generating a piece of frame omitting information in cases where the accumulated value obtained in the prediction error accumulating means is larger than a modified order threshold value Th₂ ;

moving image memory means for storing a decimated group of frames of moving images according to the frame omitting information generated in the frame omitting information generating means, the decimated group of frames being produced by omitting one or more frames from the original group of frames utilized in the prediction error accumulating means;

prediction error calculating means for calculating prediction errors of the decimated group of frames of moving images stored in the moving image memory means, each of the prediction errors being defined as a difference between a frame of moving image actually produced and a frame of moving image predicted; and

encoding means for encoding the prediction errors of the decimated group of frames calculated in the prediction error calculating means to produce prediction error codes, the amount of the prediction error codes being smaller than an upper limit, and the decimated group of frames of moving images being reproduced by decoding the prediction error codes.

13. An apparatus according to claim 12, additionally including:

weighting information generating means for generating a piece of weighting information in cases where the accumulated value obtained in the prediction error accumulating means is larger than a decimated order threshold value Th₃, the decimated order threshold value Th₃ being larger than the modified order threshold value Th₂ ;

weighting factor generating means for generating quantization weighting factors according to the weighting information generated in the weighting information generating means; and

quantizing means for quantizing the prediction errors of the decimated group of frames with the quantization weighting factors generated in the weighting factor generating means, prior to the encoding of the prediction errors in the encoding means, the amount of codes obtained by encoding the prediction errors quantized with the quantization weighting factors being smaller than the amount of codes obtained by encoding the prediction errors quantized without the quantization weighting factors.