Lossless inter-frame video coding

• US 9,210,432 B2
• Filed: 10/08/2012
• Issued: 12/08/2015
• Est. Priority Date: 10/08/2012
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google Inc. (Alphabet Inc.)
• Inventors: Gu, Qunshan, Xu, Yaowu
• Primary Examiner(s): Patel, Jay
• Assistant Examiner(s): MIKESKA, NEIL R
• Application Number:
  US13/647,092
• Publication Number:
  US 20140098855A1
• Time in Patent Office: 1,156 Days
• US Class Current: 1/1
• CPC Class Codes: H04N 19/12 Selection from among a plurality of transforms or standards, e.g. selection between discrete cosine transform [DCT] and sub-band transform or selection between H.263 and H.264
• CPC Class Codes: H04N 19/139 Analysis of motion vectors, e.g. their magnitude, direction, variance or reliability
• CPC Class Codes: H04N 19/176 the region being a block, e.g. a macroblock
• CPC Class Codes: H04N 19/61 in combination with predictive coding

Abstract

Blocks of a frame of a video stream can be encoded using lossless inter-frame prediction encoding. The compression ratio of lossless inter-frame encoding can be improved by first examining the magnitude of a motion vector used to perform inter-frame prediction. If the magnitude of the motion vector is small, the residual generated by the inter-prediction can be entropy coded and included in the encoded bitstream. If the magnitude of the motion vector is greater than a threshold, the residual can be encoded using lossy encoding and then decoded and subtracted from the original residual. The resulting difference residual is encoded using lossless encoding and included in the output bitstream along with the lossy encoded residual, permitting the block to be decoded with no loss at a decoder.

Claims

1. A method for encoding a digital video stream including a plurality of frames, the method comprising:

selecting a block of a frame of the plurality of frames, the block being one of a plurality of blocks of the frame;

performing, using a processor, inter-prediction on the block using an inter-frame prediction motion vector to calculate a first residual block by;

generating a reference block using the inter-frame prediction motion vector; and

obtaining a difference between the block and the reference block as the first residual block;

selecting one of a first coding process or a second coding process to encode the first residual block based on a comparison of a magnitude of the inter-frame prediction motion vector to a threshold value;

if the first coding process is selected, performing the first coding process as a lossless encoding process by;

entropy coding the first residual block; and

including the entropy coded first residual block in an encoded video bitstream; and

if the second coding process is selected, performing the second coding process as a lossless encoding process by;

encoding the first residual block to form an encoded residual block using a lossy encoding process;

entropy coding the encoded residual block to include in the encoded video bitstream;

decoding the encoded residual block to form a second residual block;

subtracting the second residual block from the first residual block to form a difference residual;

entropy coding the difference residual block to form an encoded difference residual block; and

including the encoded difference residual block in the encoded video bitstream.

2. The method of claim 1 wherein selecting the one of the first coding process or the second coding process comprises:

selecting the first coding process when an absolute value of the magnitude is less than or equal to the threshold value; and

selecting the second coding process when the absolute value of the magnitude is greater than the threshold value.

3. The method of claim 1 wherein encoding the first residual block comprises:

transforming the first residual block to form a transformed residual block; and

quantizing the transformed residual block using a quantization value to form a quantized residual block; and

wherein decoding the encoded residual block comprises;

de-quantizing the transformed residual block to form a de-quantized second residual block; and

inverse transforming the de-quantized second residual block to form the second residual block.

4. The method of claim 3 wherein the quantization value is a block-specific quantization value.

5. The method of claim 1, further comprising:

selecting a second block of the plurality of blocks of the frame;

performing intra-prediction on the second block to calculate a second residual block;

performing the second coding process using the second residual block by;

encoding the second residual block to form a second encoded residual block;

entropy encoding the second encoded residual block to include in the encoded video bitstream;

decoding the second encoded residual block to form a third residual block;

subtracting the third residual block from the second residual block to form a second difference residual;

entropy coding the second difference residual block to form a second encoded difference residual block; and

including the second encoded difference residual block in the encoded video bitstream.

6. The method of claim 1, further comprising:

selecting a second block of the plurality of blocks of the frame;

performing inter-prediction on the second block using a second inter-frame prediction motion vector to calculate a second residual block;

selecting one of the first coding process or the second coding process to encode the second block based on a comparison of a magnitude of the second inter-frame prediction motion vector to the threshold value;

if the first coding process is selected, performing the first coding process by;

entropy coding the second residual block; and

including the entropy coded second residual block in the encoded video bitstream; and

if the second coding process is selected, performing the second coding process by;

encoding the second residual block to form a second encoded residual block;

entropy coding the second encoded residual block to include in the encoded video bitstream;

decoding the second encoded residual block to form a third residual block;

subtracting the third residual block from the second residual block to form a second difference residual;

entropy coding the second difference residual block to form a second encoded difference residual block; and

including the encoded second difference residual block in the encoded video bitstream.

7. The method of claim 6 wherein encoding the first residual block comprises:

transforming the first residual block to form a first transformed residual block; and

quantizing the first transformed residual block using a first quantization value to form a first quantized residual block; and

wherein encoding the second residual block comprises;

transforming the second residual block to form a second transformed residual block; and

quantizing the second transformed residual block using a second quantization value to form a second quantized residual block.

8. The method of claim 7 wherein the first quantization value and the second quantization value are different values.

9. The method of claim 7 wherein the first quantization value and the second quantization value are equal values.

10. The method of claim 1, further comprising:

adding bits to a header in the encoded video bitstream to indicate that the encoded difference residual block is included in the encoded video bitstream.

11. An apparatus for encoding a digital video stream including a plurality of frames, the apparatus comprising:

a memory; and

a processor configured to execute instructions stored in the memory to;

select a block of a frame of the plurality of frames, the block being one of a plurality of blocks of the frame;

perform inter-prediction on the block using an inter-frame prediction motion vector to calculate a first residual block by;

generating a reference block using the inter-frame prediction motion vector; and

obtaining a difference between the block and the reference block as the first residual block;

select one of a first coding process or a second coding process to encode the first residual block based on a comparison of a magnitude of the inter-frame prediction motion vector to a threshold value;

if the first coding process is selected, perform the first coding process as a lossless encoding process by;

entropy coding the first residual block; and

including the entropy coded first residual block in an encoded video bitstream; and

if the second coding process is selected, perform the second coding process as a lossless encoding process by;

encoding the first residual block to form an encoded residual block using a lossy encoding process;

entropy coding the encoded residual block to include in the encoded video bitstream;

decoding the encoded residual block to form a second residual block;

subtracting the second residual block from the first residual block to form a difference residual;

entropy coding the difference residual block to form an encoded difference residual block; and

including the encoded difference residual block in the encoded video bitstream.

12. The apparatus of claim 11 wherein the processor is configured to select the one of the first coding process or the second coding process by:

selecting the first coding process when an absolute value of the magnitude is less than or equal to the threshold value; and

selecting the second coding process when the absolute value of the magnitude is greater than the threshold value.

13. The apparatus of claim 11 wherein the processor is configured to encode the first residual block using a quantization step and wherein the processor is configured to select an optimized quantization value for the quantization step.

14. A method for decoding an encoded video bitstream including a plurality of encoded frames, the method comprising:

receiving an encoded residual block of a frame of the plurality of encoded frames, the encoded residual block being one of a plurality of encoded residual blocks of the frame;

determining whether an encoded difference residual block is associated with the encoded residual block;

determining whether the encoded residual block was inter-predicted;

when the encoded residual block was inter-predicted and no encoded difference residual block is associated with the encoded residual block, reconstructing a first block of a frame of a digital video stream by;

entropy decoding the encoded residual block to form a residual block;

generating a reference block using an inter-frame prediction motion vector; and

adding the reference block to the residual block to reconstruct the first block of the frame of the digital video stream;

when the encoded residual block was inter-predicted and the encoded difference residual block is associated with the encoded residual block, reconstructing the first block of the frame of the digital video stream by;

decoding the encoded residual block to form a second residual block by;

entropy decoding the encoded residual block to form a first quantized residual block;

dequantizing the first quantized residual block to form a first transformed residual block; and

inverse transforming the first transformed residual block to form the second residual block;

entropy decoding the encoded difference residual block to form a difference residual block;

adding the second residual block to the difference residual block;

generating the reference block using the inter-frame prediction motion vector; and

adding the reference block to a sum of the second residual block and the difference residual block to reconstruct the first block of the frame of the digital video stream.

15. The method of claim 14 wherein, when the encoded residual block was intra-predicted, the method comprises:

decoding the encoded residual block to form the second residual block by;

entropy decoding the encoded residual block to form a first quantized residual block;

dequantizing the first quantized residual block to form a first transformed residual block; and

inverse transforming the first transformed residual block to form the second residual block;

entropy decoding the encoded difference residual block to form the difference residual block;

adding the second residual block to the difference residual block;

generating the reference block using intra-prediction; and

adding the reference block to the sum of the second residual block and the difference residual block to reconstruct the first block of the frame of the digital video stream.

16. The method of claim 14, further comprising:

receiving at least one bit from a header included in the encoded video bitstream indicating whether the encoded reference difference residual block is associated with the encoded residual block.