Microsoft Technology Licensing, LLC

Xiulian Peng

Jizheng Xu

Feng Wu

Yan Lu

Sriram Srinivasan

Vinod Prakash

XiangYu Kong

Fei Liang

Ji-Zheng Xu

Non-transform blocks of video content may be coded by selectively changing a scanning order. In some implementations, an encoder or decoder may perform a scanning process on a block according to a scanning order. When the block has not been transformed, the encoder or decoder may selectively change the scanning order to a different order. For example, if the block is associated with a particular type of prediction, the scanning order may be changed to a different order. In another example, if the non-transform block has a size that is smaller than a predetermined size, the scanning order may be changed to a different order.

Scanning orders for non-transform coding

Disclosed in some examples are methods, systems, and machine-readable mediums which determine jitter buffer delay by inputting jitter buffer and currently observed network status information to a machine learned model that is trained using a reinforcement learning (RL) method. The model maps these inputs to an action to compress, stretch, or hold the jitter buffer delay, which is used by a recipient computing device to optimize the jitter buffer delay. The model may be trained using a simulator that uses network traces of past real streaming sessions (e.g., communication sessions) of users. By training the model through reinforcement learning, the model learns to make better decisions through reinforcement in the form of reward signals that reflect the performance of each decision.

Reinforcement learning for jitter buffer control

Non-transform blocks of video content may be coded by selectively changing a scanning order. in some implementations, an encoder or decoder may perform a scanning process on a block according to a scanning order. When the block has not been transformed, the encoder or decoder may selectively change the scanning order to a different order. For example, if the block is associated with a particular type of prediction, the scanning order may be changed to a different order. In another example, if the non-transform block has a size that is smaller than a predetermined size, the scanning order may be changed to a different order.

Some implementations can compress a digital image to create a compressed image corresponding to the digital image. The digital image can be divided into a plurality of coding units and a determination can be made whether to divide each coding unit of the plurality of coding units into smaller coding units/prediction units based on (i) a range of pixel values in each coding unit and/or (ii) a number of bits to encode the coding unit. Rate distortion optimization can be performed based on a size of each of the prediction units. A determination can be made whether to perform a transform for each of the transform units based on a size of each prediction unit and/or a spatial frequency measure associated with each transform unit.

Video encoder for images

Images can be coded based on filters defined by filter coefficients or weights assigned to previously decoded pixel values of nearby lines. The filter coefficients can be selected based on evaluation of a set of predetermined filters, or filters can be adaptively generated and then evaluated. The filters generally are based on pixel values in previous rows or columns, but not values in the current row or column. In some examples, filters are adaptively generated pixel by pixel, or portions of previous rows, columns, or previously decoded portions of the current row or columns are used to provide line templates. A filter is generally selected to reduce coding error based on a least square difference between the current value and a prediction.

Growing community of inventors

Beijing, China

Xiulian Peng