Cisco Technology, Inc.

Tandberg Telecom As

Fileflow As

Cisco Systems International S.a.r.l.

Arild Fuldseth

Gisle Bjontegaard

Steinar Midtskogen

Gisle Bjøntegaard

Geirr Ivarsson Leistad

Tom-Ivar Johansen

Rolf Michelsen

Nils-Johan Pedersen

Andreas Kaljord

John Markus Lervik

Gisle BJØNTEGAARD

Presented herein are a variety of palette mode encoding and decoding techniques that can achieve further compression benefits. The techniques can be generalized to use arbitrary block partitions instead of rows, for instance columns of identical indices, or quadrants of identical indices.

Representing color indices by use of constant partitions

A target sample x(i,j) of a two-dimensional array of reconstructed samples is filtered based on values of samples in a neighboring region of the target sample to produce a two-dimensional array of modified reconstructed samples, according to the equation: y(i,j)=round(x(i,j)+g(&#x3a3;a(m,n)&#x192;(x(i,j)&#x2212;b(m,n)&#xd7;(m,n)))), where y(i,j) is a modified target sample value, R is the neighboring region of the target sample, a(m,n) and b(m,n) are real-valued coefficients, round(x) is a function that maps the value x to an integer value in the range [0,2&#x2212;1], B is the number of bits representing each sample of the two-dimensional array of modified reconstructed samples, f(x) and g(x) are functions, wherein (a) f(x) is a non-linear function, or (b) g(x) is a non-linear function and both a width and a height of R is more than one sample.

Generalized filter for removing video compression artifacts

Techniques for direct mode video coding/decoding techniques. According to one embodiment, various blocks in B frames may use motion vectors from different reference frames (P frames or B frames), according to what is referred to herein as a patchwork motion field. Different blocks of a given frame may use motion vectors inherited from different frames for purposes of computing a predicted block for the given frame.

Motion compensation using a patchwork motion field

A target sample x(i,j) of a two-dimensional array of reconstructed samples is filtered based on values of samples in a neighboring region of the target sample to produce a two-dimensional array of modified reconstructed samples, according to the equation: y(i,j)=round(x(i,j)+g(&#x3a3;a(m,n)&#x192;(x(i,j)&#x2212;b(m,n)x(m,n)))), where y(i,j) is a modified target sample value, R is the neighboring region of the target sample, a(m,n) and b(m,n) are real-valued coefficients, round(x) is a function that maps the value x to an integer value in the range [0,2&#x2212;1], B is the number of bits representing each sample of the two-dimensional array of modified reconstructed samples, &#x192;(x) and g(x) are functions, wherein (a) &#x192;(x) is a non-linear function, or (b) g(x) is a non-linear function and both a width and a height of R is more than one sample.

Video transform coefficient encoding and decoding methods are provided. A two-dimensional array of quantized transform coefficients are obtained for a block of a video frame. The two-dimensional array of quantized transform coefficients are converted to a one-dimensional array of integer samples using a scan pattern, wherein each sample is represented with a level that is an absolute value of the sample and a sign bit if the level is greater than zero. The one-dimensional array of samples is converted to a bit-stream by processing the one-dimensional array of samples in sequential order. A switch is adaptively made between multiple coding modes, where each coding mode is characterized by how consecutive coefficients are jointly coded as a single event. For each mode, a switch may be adaptively made between various variable length code (VLC) tables.

Transform coefficient coding using level-mode and run-mode

A target sample x(i,j) of a two-dimensional array of reconstructed samples is filtered based on values of samples in a neighboring region of the target sample to produce a two-dimensional array of modified reconstructed samples, according to the equation: y(i,j)=round (x(i, j)+g(&#x3a3;a(m, n)f(x(i,j)&#x2212;b(m,n)x(m,n)))), where y(i,j) is a modified target sample value, R is the neighboring region of the target sample, a(m,n) and b(m,n) are real-valued coefficients, round(x) is a function that maps the value x to an integer value in the range [0,2-1], B is the number of bits representing each sample of the two-dimensional array of modified reconstructed samples, f(x) and g(x) are functions, wherein (a) f(x) is a non-linear function, or (b) g(x) is a non-linear function and both a width and a height of R is more than one sample.

A block of size N&#xd7;M of a video frame can be either split into smaller blocks in n different ways (e.g. horizontal, vertical, quad, etc.), or alternatively it can be coded using one out of m different coding modes (e.g. intra, inter, bi-prediction, etc.). Since splitting into smaller sub-blocks and coding the N&#xd7;M block in one of the m possible ways are mutually exclusive, there are n+m distinct possibilities for processing the block further. The n+m possible ways of processing the N&#xd7;M blocks are assigned a code (e.g., a number) and coded with a codeword derived from a VLC table. This achieves efficient combined coding of split information and other block-level parameters.

Combined coding of split information and other block-level parameters for video coding/decoding

A video decoding method is provided. The method includes receiving an encoded bit stream representative of a video signal, the encoded bit stream including information indicating partitioning of a frame of the video signal into blocks such that at least one rectangular block of any size occurs at any position of the frame. Information is obtained from the encoded bit stream, the information signaling at least one video coding parameter for the at least one rectangular block. The encoded bit stream of the video signal is decoded in accordance with the information for the at least one rectangular block. An analogous video encoding method is also provided.

Block-based video coding using a mixture of square and rectangular blocks

A video coding/decoding system, method and computer program product employ an integer transform matrix for transforming to/from transform coefficients and residual pixel data in moving pictures by a set of semi-orthonormal basis vectors. The basis vectors are derived from conventional DCT or KTL matrixes, but relaxes to some extent the requirements for orthogonality, norm equality and element size limitation. In this way improved coding efficiency and lower complexity compared to previously used integer transforms are possible.

Integer transform video compression system, method and computer program product

A video decoding method including: receiving an encoded block of data; decoding the encoded block of data to generate decoded data; and performing a sample adaptive offset on a whole or part of the decoded data, wherein a pixel B to be modified as part of the sample adaptive offset is adjacent to pixels A and C, and the pixel B is modified as part of the sample adaptive offset if (B&#x2266;A and B&#x2266;C) or (B&#x2267;A and B&#x2267;C).

Digital video compression system, method and computer readable medium

A method including: obtaining video information at a video decoder apparatus, the video information including largest coding units of video data and filtering information, each of the largest coding units having a common size; obtaining, at the video decoder apparatus, an adaptive loop filter on/off indicator for each of the largest coding units, wherein each of the largest coding units includes a respective adaptive loop filter on/off indicator; and performing, by the video decoder apparatus, adaptive loop filtering to the largest coding units if the respective adaptive on/off indicators are on.

Video compression using an adaptive loop filter

In video encoding it is common to encode the image data to remove redundancies in the information to be transmitted. While the pixel data is usually arranged in pixel blocks, the blocks can be arranged in one or more groups of N&#xd7;M blocks called tiles. The tiles avoid the need to send header information on a tile-by-tile or block-by-block basis, and simplifies parallel processing of the tiles. Bits from respective tiles may then be reformatted to recreate bits according to a raster-scan direction. This enables the decoder to receive the bits in a regular raster-scan format, but also have the ability to decode the tiles once the bits are reformatted. By partitioning an image into tiles of size N&#xd7;M, it is possible to further exploit the intra-frame correspondence of images in a vertical direction as well as horizontal direction since the tiles need not destroy as many dependencies between blocks in a tile as if the blocks where organized in slices or slice groups.

Video encoder/decoder, method and computer program product that process tiles of video data

A video decoder, method and computer program product allow for processing of a video frame encoded in rectangular tiles. An interface receives a bit stream in tile order within a video frame that was encoded into rectangular tiles. A processor decodes the video frame while respecting dependency breaks at tile boundaries; the rectangular tiles include an integer number of two-dimensional blocks of pixels. A tile shape is defined by N&#xd7;M two-dimensional blocks of pixels, respective values of N and M need not be identical for each of the rectangular tiles, and information regarding tile shape for each tile being conveyed from an encoder to the decoder. The decoder determines N and M for each tile from the information, and tiles have dependency breaks therebetween.

A video decoding method, system and computer program product perform inverse transforming of transform coefficients for a video image. The inverse transforming is performed with a processing circuit on a first block of transform coefficients by a transform matrix having a plurality of vectors. The coefficients are transformed into a corresponding second block of residual pixel values for the video image. The transform matrix consists of integer elements, and includes a 4 point transform as whole or partial vectors of the transform matrix. The first row of the 4 point transform includes elements a a a a. The second row includes b c &#x2212;c &#x2212;b. The third row includes a &#x2212;a &#x2212;a a. The fourth row includes c &#x2212;b b &#x2212;c, wherein a, b and c comply with |b+c&#x2212;(a+a)|/(a+a)&#x3c;0.003, b/c being in an inclusive range of 2.1 through 2.4, and each of a, b and c being integer.

Growing community of inventors

Lysaker, Norway

Arild Fuldseth