Apple Inc.

Krishnakanth Rapaka

Alexandros Michael Tourapis

David William Singer

Hao Pan

Munehiro Nakazato

Jim Chen Chou

Dominik Mehlem

Jungsun Kim

Hsi-Jung Wu

Haitao Guo

Sorin Constantin Cismas

Jun Xin

Mei Guo

Xiaohua Yang

Ganesh G Yadav

Jiandong Shen

Linfeng Guo

Wei Li

Ke Chen

Haiyan He

Jaehong Chon

Francesco Iacopino

Ye Cong

Xiang Fu

Xu Zhao

Techniques for encoding video with motion compensation include a compressed bitstream syntax that includes a list of all motion prediction reference frames without distinguishing between short-term reference frame and long-term reference frames. The list of reference frames may be provided in a slice header and may apply to encoded data video data within the corresponding slice. The list may be prefaced with a single number indicating the total number of reference frames. In an aspect delta POC reference numbers may be encoded with a flag indicating the sign of the delta POC when the absolute value of the POC is not equal to zero. In another aspect, a flag may be encoded for every reference frame indicating if POC information should be used when scaling prediction references, and a weighting parameter may be included when POC information should be used.

Reference picture management and list construction

Techniques are disclosed for deriving prediction pixel blocks for use in intra-coding video and combined inter- and intra-coding video. In a first aspect, the techniques may include deriving value(s) for pixel location(s) of the prediction pixel block by, when a prediction direction vector assigned to the prediction vector points to quadrants I or III of a Cartesian plane, deriving the pixel location's value from pixel values in two regions of previously-decoded pixel data intercepted by extending the prediction direction vector in two opposite directions through the pixel location. When the prediction direction vector points toward quadrants II of the Cartesian plane, deriving the pixel location's value from pixel values in one region intercepted by the prediction direction vector through the pixel location, and from a second region intercepted by a vector that is orthogonal to the prediction direction vector.

Smoothed directional and DC intra prediction

A system comprises an encoder configured to compress video data. The encoder includes an adaptive bilateral filter that uses look-up tables. The encoder may encode one or more adaptive adjustment factors to be used by a decoder to select or adjust look-up tables used to decode the compressed video data.

Adaptive bilateral filtering using look-up tables

Predictive coding techniques may include resampling of reference pictures, where various coding parameters are determined based on the resolution(s) or pixel format(s) of the prediction references. In a first aspect, lists of weights for use in weighted prediction are based on the resolution(s) of prediction references. In a second aspect, resampling filter parameters are selected based on the resolutions of prediction references. In a third aspect, deblocking filter parameters are based on the resolution(s) of prediction references.

Reference picture re-sampling

Improved video coding and decoding techniques are described, including techniques to derive quantization step sizes adaptively with quantization step size table templates. Quantization techniques described provide finer-grained control over quantization with a more flexible quantization step size especially at higher degrees of quantization. This may result in improved overall compression quality. Other coding parameters, such as in-loop filtering parameters, may be derived based on the more flexible quantization parameters.

Adaptive quantizer design for video coding

A multi-stage coding method includes receiving an input block of data for encoding and one or more previously coded samples associated with the input block. The input block is segmented into at least a first sub-region and a second sub-region. A prediction for the first sub-region is generated based on the one or more previously coded samples. Residual data for the first sub-region is obtained using the prediction for the first sub-region. A reconstruction of the first sub-region is generated using the residual data for the first sub-region and the prediction for the first sub-region. A prediction for the second sub-region is generated using the reconstruction of the first sub-region. Residual data for the second sub-region is obtained using the prediction for the second sub-region. The input block is encoded based in part on the residual data for the first region and the residual data for the second region.

Multi-stage block coding

In one implementation, a method of generating a bit stream encoding a video stream is performed by a device including one or more processors and non-transitory memory. The method includes decomposing a video stream into a plurality of frequency band video streams. The method includes determining a target bitrate and determining, for each frequency band video stream, a respective frequency band bit rate based on the target bit rate. The method includes encoding each of the plurality of frequency band video streams at its respective frequency band bit rate and transmitting, over a channel, each encoded frequency band video stream.

Multi-band rate control

In one implementation, a method of encoding an image is performed at a device including one or more processors and non-transitory memory. The method includes determining a category of a spatial portion of an image based on a relation between a plurality of thresholds associated with a plurality of quantization scaling parameters and a bit rate of the spatial portion of the image at the plurality of quantization scaling parameters. The method includes quantizing the spatial portion of the image based on the categorization.

Video encoder with quantization control

Techniques for coding and decoding video may include predicting picture regions defined by a time-varying tessellation and/or by a tessellation that varies spatially within a picture. These techniques improve decoded video quality, for example, by reducing block-based visual artifacts. Tessellation patterns may be irregular spatially to prevent alignment of some prediction region boundaries within a picture. Tessellation patterns may vary over time based on a spatial offset value, and the spatial offset value may be determined via a modulo function. Tessellation patterns may include overlapped shapes, for example when used in conjunction with overlapped block motion compensation.

Flexible block partitioning structures for image/video compression and processing

In one implementation, a method of encoding an image is performed at a device including one or more processors and non-transitory memory. The method includes receiving a first image comprising a plurality of pixels having a respective plurality of pixel locations and a respective plurality of pixel values. The method includes applying a frequency transform to a first spatial portion of the first image to generate a plurality of first frequency coefficients respectively associated with a plurality of spatial frequencies and applying the frequency transform to a second spatial portion of the first image to generate a plurality of second frequency coefficients respectively associated with the plurality of spatial frequencies. The method includes obtaining a first quantization matrix including, for each of the plurality of spatial frequencies, a respective first quantization parameter, wherein a first quantization parameter associated with a first spatial frequency of the plurality of spatial frequencies has a first ratio with a first quantization parameter associated with a second spatial frequency of the plurality of spatial frequencies, and obtaining a second quantization matrix including, for each of the plurality of spatial frequencies, a respective second quantization parameter, wherein a second quantization parameter associated with the first spatial frequency has a second ratio, different than the first ratio, with a second quantization parameter associated with the second spatial frequency. The method includes quantizing the plurality of first frequency coefficients according to the first quantization matrix to generate a plurality of first quantized frequency coefficients and quantizing the plurality of second frequency coefficients according to the second quantization matrix to generate a plurality of second quantized frequency coefficients. The method includes transmitting the plurality of first quantized frequency coefficients and the plurality of second quantized frequency coefficients.

Improved video coding techniques include deriving prediction pixel blocks for use in intra-coding video and combined inter- and intra-coding video. In a first aspect, the techniques may include deriving value(s) for pixel location(s) of the prediction pixel block by, when a prediction direction vector assigned to the prediction vector points to quadrants I or III of a Cartesian plane, deriving the pixel location's value from pixel values in two regions of previously-decoded pixel data intercepted by extending the prediction direction vector in two opposite directions through the pixel location. When the prediction direction vector points toward quadrants II of the Cartesian plane, deriving the pixel location's value from pixel values in one region intercepted by the prediction direction vector through the pixel location, and from a second region intercepted by a vector that is orthogonal to the prediction direction vector.

Growing community of inventors

San Jose, CA, United States of America

Krishnakanth Rapaka