Motion Prediction In Video Coding

Patent No. US11805267 (titled "Motion Prediction In Video Coding") was filed by Nokia Technologies Oy on May 24, 2021.

What is this patent about?

’267 is related to the field of video coding and, more specifically, to improving the efficiency of motion-compensated prediction in video codecs. Modern video codecs use techniques like motion compensation to reduce the amount of data needed to represent video, by predicting pixel values in a frame based on previously encoded frames. This involves finding similar areas in reference frames and encoding only the differences, along with motion vectors indicating the displacement.

The underlying idea behind ’267 is to reduce rounding errors that accumulate during bi-directional or multi-directional prediction. Instead of rounding each prediction signal individually before combining them, the patent proposes maintaining the prediction signals at a higher precision throughout the calculation. Only after the predictions are combined is the precision reduced to the original level, thereby minimizing the impact of rounding errors on the final prediction.

The claims of ’267 focus on a method and apparatus for encoding and decoding video. For a given block of pixels, the method involves determining two or more reference blocks using motion vectors. Each reference block is used to generate a prediction with a higher precision than the original pixel values. These predictions are then combined, and the precision of the combined prediction is reduced by bit-shifting , effectively rounding the result to the original precision.

In practice, the invention involves using interpolation filters to obtain prediction values from reference blocks. These filters operate on pixel values in the reference blocks to estimate values at fractional pixel locations, allowing for more accurate motion compensation. The key is that the intermediate results of these filtering operations, as well as the subsequent combination of predictions, are performed using a higher bit depth (i.e., higher precision) than the original video data. This reduces the quantization noise introduced by rounding at each stage.

This approach differs from prior methods that might round each prediction signal before combining them, or that might signal rounding offsets in the bitstream. By maintaining higher precision throughout the prediction calculation, ’267 avoids the need to signal rounding direction and reduces the accumulation of rounding errors, leading to improved coding efficiency without significantly increasing decoder complexity. The final bit-shift operation ensures that the output prediction has the same precision as the original video signal.

How does this patent fit in bigger picture?

Technical landscape at the time

In the early 2010s when ’267 was filed, video codecs commonly employed hybrid encoding schemes, at a time when motion compensation was typically implemented using block-based prediction. Inter prediction methods using image information from previous or later images and Intra prediction methods using image information within the same image were both prevalent. Hardware constraints made maintaining high precision throughout the prediction process non-trivial, often requiring trade-offs between computational complexity and coding efficiency.

Novelty and Inventive Step

The application was subject to a non-final rejection based on obviousness-type double patenting. Claims 1, 7, 13, 19, 25 and 31 were provisionally rejected as being unpatentable over claim 1 of U.S. Patent No. 11,019,354 and U.S. Patent No. 10,523,960. The prosecution record does not describe the technical reasoning or specific claim changes that led to allowance.

Claims

This patent contains 36 claims, with independent claims numbered 1, 7, 13, 19, 25, and 31. The independent claims are generally directed to methods, apparatuses, and computer program products for encoding and decoding blocks of pixels using motion vectors and precision adjustments. The dependent claims generally elaborate on specific aspects and implementations of the methods, apparatuses, and computer program products described in the independent claims.

Key Claim Terms New

Definitions of key terms used in the patent claims.

Term (Source)Support for SpecificationInterpretation
Combined prediction
(Claim 1, Claim 7, Claim 13, Claim 19, Claim 25, Claim 31)
“In some example embodiments prediction signals are maintained in higher accuracy until the prediction signals have been combined to obtain the bi-directional or multidirectional prediction signal. The accuracy of the bi-directional or multidirectional prediction signal can then be downshifted to an appropriate accuracy for post processing purposes.”A prediction obtained by combining the first prediction and the second prediction.
First precision
(Claim 1, Claim 7, Claim 13, Claim 19, Claim 25, Claim 31)
“According to some embodiments of the invention prediction signals are maintained in a higher precision during the prediction calculation and the precision is reduced after the two or more prediction signals have been combined with each other.”The precision of the pixel values of the current block, the first reference block, and the second reference block.
First reference block
(Claim 1, Claim 7, Claim 13, Claim 19, Claim 25, Claim 31)
“In the first phase, pixel values in a certain picture area or “block” are predicted. These pixel values can be predicted, for example, by motion compensation mechanisms, which involve finding and indicating an area in one of the previously encoded video frames (or a later coded video frame) that corresponds closely to the block being coded.”A block of pixels used as a reference for prediction, determined based on a first motion vector.
Second precision
(Claim 1, Claim 7, Claim 13, Claim 19, Claim 25, Claim 31)
“According to some embodiments of the invention prediction signals are maintained in a higher precision during the prediction calculation and the precision is reduced after the two or more prediction signals have been combined with each other.”A precision higher than the first precision, used for the first and second predictions.
Second reference block
(Claim 1, Claim 7, Claim 13, Claim 19, Claim 25, Claim 31)
“In the first phase, pixel values in a certain picture area or “block” are predicted. These pixel values can be predicted, for example, by motion compensation mechanisms, which involve finding and indicating an area in one of the previously encoded video frames (or a later coded video frame) that corresponds closely to the block being coded.”A block of pixels used as a reference for prediction, determined based on a second motion vector.

Litigation Cases New

US Latest litigation cases involving this patent.

Case NumberFiling DateTitle
1:25-cv-00523Apr 7, 2025Nokia Technologies Oy V. Acer Inc.
0:24-cv-04269Nov 25, 2024Element Television Company, Llc V. Nokia Corporation
1:23-cv-01237Oct 31, 2023Nokia Technologies Oy V. Hp, Inc.
1:23-cv-01236Oct 31, 2023Nokia Technologies Oy V. Amazon.Com, Inc.

Patent Family

Patent Family

File Wrapper

The dossier documents provide a comprehensive record of the patent's prosecution history - including filings, correspondence, and decisions made by patent offices - and are crucial for understanding the patent's legal journey and any challenges it may have faced during examination.

  • Date

    Description

  • Get instant alerts for new documents

US11805267

NOKIA TECHNOLOGIES OY
Application Number
US17328750
Filing Date
May 24, 2021
Status
Granted
Expiry Date
Jan 6, 2032
External Links
Slate, USPTO, Google Patents