Coded data generation or conversion – Digital code to digital code converters – Adaptive coding
Reexamination Certificate
1998-08-06
2001-04-10
Wamsley, Patrick (Department: 2819)
Coded data generation or conversion
Digital code to digital code converters
Adaptive coding
C341S065000, C382S239000, C382S246000
Reexamination Certificate
active
06215422
ABSTRACT:
The present invention concerns in general digital signal coding and to that end proposes a device and method for coding a digital signal by breaking the signal down into frequency sub-band signals and coding the coefficients resulting from the breakdown into sub-band signals. It also concerns a decoding method and device corresponding to the coding method and device. The present invention also proposes a device and a method for coding a digital signal which offers an adaptive allocation of transmission rate to the signal to be coded. It also concerns a decoding method and device corresponding to the coding method and device.
The aim of the coding is to compress the signal, which makes it possible to transmit, or respectively store, the digital signal while reducing the transmission time, or the transmission rate, or respectively reducing the memory space used.
The invention belongs to the field of digital signal compression with loss.
Breaking a signal down into frequency sub-band signals before compressing it is known. The breaking down consists of creating, from the signal, a set of sub-band signals which each contain a limited range of frequencies. The sub-band signals may be of different resolutions, the resolution of a sub-band signal being the number of samples per unit length used to represent that sub-band signal. In the case of a digital image signal, a frequency sub-band signal of that signal may itself be considered as an image, that is to say a bidimensional table of digital values.
It should be noted that breaking a signal down into frequency sub-band signals does not in itself create any compression, but makes it possible to decorrelate the signal so as to eliminate redundancy from it prior to the compression proper. The sub-band signals are thus coded more efficiently than the original signal.
A first known method of coding a digital signal, in this case a digital image, has three main steps. The image is first of all broken down by a transformation into frequency sub-band signals, in order to form frequency sub-band signals according to a number of resolutions. The frequency sub-band signals are next divided into blocks. Each block can be coded by two different coding modes, for example by inter-resolution prediction, or by residual block coding, if coding of the block under consideration by prediction does not give a satisfactory result.
An additional information is added to the data for coding each block, to indicate which coding mode was used to code the block under consideration.
This method has good coding accuracy, since two coding modes can be used for each block, according to a quality criterion. The coding error is therefore smaller than if a single coding mode is used. However, it is necessary to add supplementary information to indicate which coding mode was used to code each block. This supplementary information decreases the compression ratio.
A second known method of coding a digital signal, in this case a digital image, includes three main steps. The image is first of all broken down by means of a transformation into frequency sub-band signals, and then a scalar quantization of the coefficients thus obtained is performed. The quantified indices are finally coded by means of an entropic coding without loss.
This method affords a high compression ratio for the signal. However, the ratio between the compression ratio and the coding error is open to improvement.
A third known method of coding a digital signal, in this case a digital image, uses the similarity between blocks of different sub-band signals of a multi-resolution breakdown of the image.
This method makes it possible to obtain a good visual quality of the restored image and limited distortion. However, the ratio between the compression ratio and the coding error is open to improvement.
A fourth known method of coding a digital signal, in this case a digital image, divides the signal into parts and allocates a coding mode to each part in an optimum fashion, which makes it possible to adapt the transmission rate in an optimum manner. Hereinafter, it is considered that the coding of a digital signal includes two main operations: the quantification of the signal, for example a vectorial quantification, which supplies a series of symbols, and then an entropic coding of these symbols, for example a Huffman coding, which supplies a binary series.
The coding modes are composed of a quantification mode, and an entropic coding mode. The known method consists of testing all the combinations of quantification mode and entropic coding mode, for each part of the signal, and then selecting the best combination, according to a criterion, for each part of the signal.
This method entails numerous calculations, the quantity of which increases with the number of quantification modes, entropic coding modes and parts of the signal. In practice, it is necessary to reduce one of these numbers, for example the number of parts of the signal, in order to reduce the calculation time. This reduction is achieved to the detriment of the adaptability of the transmission rate to the signal to be coded. The ratio of compression to distortion is consequently open to improvement.
The present invention aims to remedy the drawbacks of the prior art by providing a device and method for compressing a digital signal which offers a high ratio of compression to distortion.
To that end, the invention firstly proposes a digital signal coding method having a step of analysing the digital signal into a plurality of frequency sub-band signals distributed according to at least two different frequency bands, at least one first sub-band signal having a lower frequency and at least one second sub-band signal having a higher frequency,
characterised in that it has, for each second sub-band signal, the steps of:
dividing the second sub-band signal into blocks,
selecting a coding mode for each block to be coded from amongst a predetermined set of coding modes, according to a selection criterion,
associating an indicator with each block to be coded, the value of which is a function of the coding mode selected for the block under consideration,
grouping the indicators into indicator words, according to a grouping criterion, and
entropic coding of the indicator words.
Correlatively, the invention proposes a digital signal coding device having means of analysing the digital signal into a plurality of frequency sub-band signals distributed according to at least two different frequency bands, at least one first sub-band signal having a lower frequency and at least one second sub-band signal having a higher frequency,
characterised in that it has:
means of dividing each second sub-band signal into blocks,
means of selecting a coding mode for each block to be coded from amongst a predetermined set of coding modes, according to a selection criterion,
means of associating an indicator with each block to be coded, the value of which is a function of the coding mode selected for the block under consideration,
means of grouping the indicators into indicator words, according to a grouping criterion, and
means of entropic coding of the indicator words.
The digital signal coding device and method according to the invention have a high ratio of compression ratio over coding error. This is because the use of a number of coding modes makes it possible to choose the most appropriate coding mode for each block, which reduces the overall coding error.
An indicator for each block is necessary for the decoding in order to indicate which coding mode was used. There are therefore as many indicators as blocks, which reduces the compression ratio. Grouping of the indicators into words, the said words then being coded, makes it possible to limit the compression ratio reduction.
Thus the invention has overall a ratio of compression ratio over coding error which is high.
According to a preferred characteristic, the grouping criterion consists of grouping together indicators of blocks selected from sub-band signals in order to correspond to the same part of the digital sign
Amonou Isabelle
Dierieck Claude
Henry Felix
Canon Kabushiki Kaisha
Fitzpatrick ,Cella, Harper & Scinto
Wamsley Patrick
LandOfFree
Digital signal huffman coding with division of frequency... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Digital signal huffman coding with division of frequency..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Digital signal huffman coding with division of frequency... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2446801