Intel Corporation

Rohit Agarwal

Joseph N Romriell

Michael J Keith

Stuart J Golin

Thomas E Walsh

Michael J Gutmann

In block based video compression, a frame is divided into blocks which define a tiling pattern. The tiling pattern is varied from frame-to-frame to prevent an accumulation of errors which tend to appear at tile edges and can increase over time when using block-based compression. In a preferred embodiment, a normal frame is padded by a border all around the normal frame size. The padding is operable to extend any blocks around the periphery of the image frame which might be smaller in dimension than the standard blocks, such as those within the interior of the frame, such that they can be treated by the block-based compression systems as full size blocks.

Temporal tile staggering for block based video compression

A set different dither matrices is employed in a repetitive manner to accomplish elimination of contouring while at the same time not adding typical 'dither graininess' to a digital video image. The matrices are specially chosen to result in a time-averaged zero dithering bias when applied frame-by-frame in a long sequence of frames.

Temporally dissolved dithering

Two or more run-val mapping tables are selected to encode different subsets of video signals of a sequence of video signals. Each run-val mapping table is selected to map a particular set of run-val pairs to a corresponding set of run-val codes. Each run-val pair comprises a run and a vat, the run corresponding to a length of a run of video signals having value zero and the val corresponding to a video signal having a non-zero value. Each run-val mapping table is used to generate a different part of an encoded bitstream for the sequence of video signals. The encoded bitstream is then decoded using the two or more selected run-val mapping tables. In a preferred embodiment, the selected run-val mapping tables are explicitly identified in the bitstream.

Encoding/decoding video signals using multiple run-val mapping tables

The efficiency of variable-length (VL) encoding data using a defined VL encoding table is characterized and one or more changes to one or more entries of the defined VL encoding table are determined that increase the efficiency of encoding the data. An updated VL encoding table is generated from the defined VL encoding table based on the changes. VL codes for the data are generated using the updated VL encoding table, and an encoded bitstream is generated from VL codes, wherein the encoded bitstream explicitly identifies the changes. For decoding, the changes are extracted from the encoded bitstream and an updated VL decoding table is generated from a defined VL decoding table corresponding to the defined VL encoding table, based on the changes. The VL codes in the encoded bitstream are decoded using the updated VL decoding table to generate a set of decoded data.

Variable-length encoding using code swapping

One or more encoding parameters are controlled during the encoding of image data based on a decode processing rate. By taking decode processing into account during encoding, a decoding system will be able to keep up with the demands for decoding the resulting encoded bitstream. In one embodiment, encoding parameters used for video encoding may be changed iteratively to achieve a specified target decode processing rate, which can change from frame to frame.

Encoding image data for decode rate control

Sequential digital video frames are divided into two interleaved fields comprising even-numbered horizontal lines of pixels and odd-numbered lines horizontal lines of pixels respectively. The 'even' field of a first standard frame is combined with the 'odd' field of a second standard frame to form a first pseudo-interlaced frame of data for display; then the same odd field is combined with the even field of a third standard frame to form a second pseudo-interlaced frame of data for display, and so on throughout a continuing digital-video message. The data transmission rate needed is thus reduced by nearly a factor of two, or conversely, for a given data rate, nearly twice as much video information can be transmitted in the same time interval.

Pseudo interlacing in digital video

Two or more run-val mapping tables are selected to encode different subsets of video signals of a sequence of video signals. Each run-val mapping table is selected to map a particular set of run-val pairs to a corresponding set of run-val codes. Each run-val pair comprises a run and a val, the run corresponding to a length of a run of video signals having value zero and the val corresponding to a video signal having a non-zero value. Each run-val mapping table is used to generate a different part of an encoded bitstream for the sequence of video signals. The encoded bitstream is then decoded using the two or more selected run-val mapping tables. In a preferred embodiment, the selected run-val mapping tables are explicitly identified in the bitstream.

Video signals are encoded and decoded using a set of quantization tables that is generated from a base matrix and a scale matrix, which are both explicitly encoded into the encoded video bitstream. The sets of quantization tables may be changed as often as needed as long as the new base and scale matrices are explicitly encoding into the bitstream.

Encoding/decoding video signals using quantization tables based on

Video signals are encoded using motion compensation based on motion vectors generated using multi-phase motion estimation. In a first phase, a first subset of video signals of each block is compared to the reference frame at locations corresponding to a first set of motion vectors. In a second phase, a second subset of video signals of each block is compared to the reference frame at locations corresponding to a second set of motion vectors, where the resolution of the first subset of video signals in the spatial domain is lower than the resolution of the second subset of video signals. In an alternative embodiment, in the first phase, each block is compared to the reference frame at locations corresponding to a first set of motion vectors. In the second phase, each block is compared to the reference frame at locations corresponding to a second set of motion vectors, where the second set of motion vectors is based on at least two best matches from the first phase.

Encoding video signals using multi-phase motion estimation

A heterogeneous multimedia stream has interleaved samples of two or more different substream types. For example, a heterogeneous video stream may have interleaved intra, predicted, and bi-directional frames. In addition to a global bit bank, a different bit bank is maintained for each different frame type. A global quantization level is selected for each frame based on the global bit bank and the bit bank for the corresponding frame type. The frames are encoded using the selected global quantization levels. In a preferred embodiment, the number of bits available for encoding frames differs based on the frame type. In addition, any unused bits from encoding the current frame are shifted from the bit bank corresponding to the current frame type to a bit bank of a different frame type.

Interleaved bitrate control for heterogeneous data streams

The video encoding scheme used by a video encoder is selected from a variety of different possible video encoding schemes having different decode rates to generate an encoded video bitstream to achieve a desired decode rate in a specified decoding environment. The video encoder allows a user to select different characteristics of the video encoding scheme to generate the encoded video bitstream for a specific decoding platform.

Encoding images using decode rate control

Video signals are run-length encoded using a scan pattern and encoded video signals are run-length decoded using the scan pattern, where the scan pattern is explicitly encoded into the encoded bitstream. In a preferred embodiment, the scan pattern is generated by applying a constrained sorting rule to energy measures corresponding to the video signals to be encoded. The scan pattern used for encoding can be changed as often as needed where each new scan pattern is explicitly encoded into the bitstream for use in decoding.

Run-length encoding/decoding video signals using scan patterns

A video system is disclosed that provides fast communication packet loss recovery. A source node generates a series of encoded video flames for a series of digitally sampled video flames by encoding each digitally sampled video frame relative to either an immediately preceding digitally sampled video frame or a previous digitally sampled video frame specified by a lost frame message received over the communication link from a destination node. The destination node receives the encoded video frames over the communication link and transfers the lost frame message to the source node if one of the encoded video frames is lost or delayed. The destination node generates the decoded version of each encoded video frame relative to either a decoded version of an immediately preceding encoded video frame or a decoded version of a last successfully received encoded video frame.

Growing community of inventors

Hillsboro, OR, United States of America

Rohit Agarwal