Coding methods

2012-10 Multilevel Coding - SOLD

This method enables very efficient coding of different data types in the same input data. The method yields a very high compression ratio for data that has two or more levels, or for data where there is a big difference between data values and where there are small non-visible errors. Moreover, the method offers easy scaling of quality (bitrate reduction) by quantizing the values, yet not causing a large objective or subjective error.

2013-07 Extrapolation Method - SOLD

This patented coding method provides very efficient encoding for certain types of image data, video data, audio data and many other types of data. The method offers a very high compression ratio for data that has a clear structure that can be described easily with a few parameters, by utilizing an extrapolation method which makes it possible to have considerably less output values than there are input values. The method enables lossless or lossy reconstruction in a corresponding decoder, and it is beneficially used with other methods for optimization purposes, e.g. in conjunction with using RD-value for lossy reconstructed data, or in conjunction with used bits for lossless reconstructed data.

The method offers easy scaling of quality (bitrate reduction) by quantizing the values, without resulting in large objective or subjective errors. If prediction from surrounding block or packet data values is not used for the data extrapolation of data block or data packet, then the invention has a simpler parallel processing implementation available, which reduces error propagation when data is first encoded and subsequently decoded.

The method provides a technical benefit of a reduction in data size, which has technical implications in use of data storage space and energy dissipation and communication bandwidth utilization, when the encoded data is stored and/or communicated.

2013-07 Interpolation Method - SOLD

This patented coding method provides very efficient encoding for certain types of image data, video data, audio data, graphics, text, ASCII or Unicode character data, binary block data or data packets. This method offers a very high compression ratio for data that has a lot of variation, but whose frequency content is not the highest possible. Such data can be described easily with a small amount of values and/or parameters, together with an associated selected interpolation method.

The method enables both lossless and lossy reconstruction of encoded data, and it is beneficially used with other methods to optimize for example RD-values for lossy reconstructed data, or bits used for lossless reconstructed data.

The method offers easy scaling of quality (bitrate reduction) by quantizing the values, yet does not cause large objective or subjective errors in reconstructed data. If prediction from surrounding block or packet data values is not used for the data interpolation of a given block or packet, then the given data packet or the data block is fully independent on other data blocks or data packets, and simpler parallel processing implementations can be executed that reduce error propagation during decoding.

2014-06 Efficient coding of block-internal changes - PATENT FAMILY FOR SALE

The present invention is aimed at achieving a better compression ratio in case of data communication when a sequence of data blocks and/or data packets is communicated from one device to another device. The data blocks and/or data packets communicated later in the sequence are often only slightly changed in comparison to data blocks and/or data packets that have been communicated earlier. Most of the elements (namely, contents of a data block) inside the changed data blocks/packets, however, are unchanged in comparison to the contents of earlier data blocks/packets. When such a sequence of data is compressed using conventional encoders, often only a small compression ratio is achieved.

The method pursuant to this invention identifies the partially changed data blocks/packets, and codes only the changed elements while employing merely symbol or bit indicating an absence of change in the unchanged elements relative to corresponding elements in a reference data block/packet, subject to comparison.

Examples of elements can include, for example, base pairs of DNA or RNA in genomic data.

2016-07 Coding method employing differing quantization - PATENT FAMILY FOR SALE

Encoding method wherein different quantization strengths are employed for a same value of the quality parameter for at least two data portions of the input data. The method comprises selecting quantization strength portion-by-portion and assembling related information and transferring to decoder portion-by-portion.

The invention provides a cost-efficient method of changing quantization strengths e.g. block-by-block, when using various encoding methods to encode the input data, thus potentially achieving an improved compression ration or an improved quality of the encoded data.