Pulse or digital communications – Bandwidth reduction or expansion – Television or motion video signal
Reexamination Certificate
1999-08-05
2004-08-17
Chin, Stephen (Department: 2634)
Pulse or digital communications
Bandwidth reduction or expansion
Television or motion video signal
C375S240190, C382S248000
Reexamination Certificate
active
06778608
ABSTRACT:
The present invention concerns a method and a device for the geometric transcoding of a compressed digital signal. It also concerns a coding method and device on the one hand, and a decoding method and device on the other hand, associated with the geometric transcoding method and device.
The technical field of the invention is that of methods for manipulating compressed digital data.
Geometric transcoding is a transformation of one compressed digital data file into another compressed digital data file which, once decoded, represents a digital signal which has undergone a geometric transformation with respect to the initial coded digital signal in the initial compressed data file.
The geometric transformation of a mono- or multi-dimensional digital signal consists of changing the order of the samples of digital signal in one or more directions of the digital signal.
In practice, in the case for example of an image, axial or central symmetry can be produced, or a rotation of the image by manipulating the digital signal representing the image.
It is particularly advantageous to be able to carry out this manipulation and transformation of a coded digital signal without having to compress and then once again compress the file containing the coded digital signal, in order notably to reduce the number of calculations and the memory space necessary for a decompression of the compressed digital data file.
Techniques are known for manipulating digital data in a compressed form, which have been coded by a coding method based on a discrete cosine transform (DCT). Such techniques are described for example in the article “A new family of algorithms for manipulating compressed images” by Brian C. Smith, Lawrence A. Rowe, IEEE Transactions on Computer Graphics and Applications, September, 1993, and in U.S. Pat. No. 5,751,865 (Felice A. Micco, Martin E. Banton).
However, the techniques of compressing a digital signal based on a discrete cosine transform are less efficient in terms of compression than the compression techniques based on a spectral breakdown of the digital signal into frequency sub-bands.
The present information aims to remedy the drawbacks of the prior art by proposing a method and device for transcoding a compressed data file by a compression technique using a spectral breakdown of the digital signal.
To this end, the invention concerns a method for the geometric transcoding of a compressed data file containing a digital signal of dimension N coded by a coding method including at least one step of spectral breakdown into frequency sub-bands of the digital signal.
According to the invention, this transcoding method is characterised in that it includes the following steps:
extraction of symbols associated with the coefficients of the frequency sub-bands in each direction of the digital signal;
application of a geometric transformation to said symbols;
updating of N indicator or indicators representing a normal or reversed order of the symbols respectively in N direction or directions of the digital signal; and
reconstitution of the coded digital signal by reversal of the extraction step
Correlatively, the invention proposes a device for the geometric transcoding of a compressed data file containing a digital signal of dimension N coded by a coding method including at least one step of spectral breakdown into frequency sub-bands of the digital signal, characterised in that it has:
means of extracting symbols associated with the coefficients of the frequency sub-bands in each direction of the digital signal;
means of applying a geometric transformation to said symbols;
means of updating N indicator or indicators representing a normal or reversed order of the symbols respectively in N direction or directions of the digital signal; and
means of reconstituting the coded digital signal by reversal of the extraction step.
The geometric transcoding method and device make it possible to effect a geometric transformation of a coded digital signal on the basis of a spectral breakdown without completely decompressing the digital signal since the precaution is taken of modifying an indicator intended to give information about the reversal or not of the frequency sub-band symbols, in each direction of the digital signal. This indicator makes it possible to adapt the subsequent decoding method, notably the spectral reconstruction of the digital signal as a function of the value of this indicator.
In addition, this transcoding method avoids making modifications to the value of the symbols associated with the coefficients of the frequency sub-bands of the digital signal, without any increase in the calculation quantity necessary at the time of decoding.
According to one advantageous characteristic of the invention, the extraction step includes a reading of the coded digital signal, the signals being entropic codes associated respectively with the coefficients of the frequency sub-bands obtained by spectral breakdown.
This embodiment is particularly advantageous since it requires a minimal decompression of the compressed data file, the geometric transformation being applied directly to the entropic codes contained in the compressed data file, referred to as prefix codes. It is, however, well suited only when the coding of the digital signal uses an entropic coding, of the Huffman coding type, which associates an entropic coding with each coefficient of the signal of the frequency sub-bands.
According to a preferred version of the invention, which minimises the decompression of the file and is applied for a large number of conventional coding methods, using a scalar quantization of the coefficients of the signal of the sub-bands, the extraction step includes an entropic decoding of the coded digital signal, the symbols being quantization symbols associated respectively with the coefficients of the frequency sub-bands obtained by spectral breakdown.
Alternatively, according to another preferred version, which avoids the complete decompression of the file, and notably the spectral recomposition of the digital signal, the extraction step includes a dequantization of the coded digital signal, the symbols being the dequantized coefficients of the frequency sub-bands obtained by spectral breakdown.
This embodiment is particularly well suited when the coding method uses a vector quantization of the digital signal broken down into frequency sub-bands.
According to a preferred version of the invention, the indicator is a supplementary bit recorded in the compressed data file, having an initial value representing a normal order of the coefficients of the frequency sub-bands in a direction associated with said indicator.
Such an indicator is particularly advantageous in terms of memory space used in the compressed data file. It is also very easily updated from its initial value, fixed for example at the time of coding of the digital signal which has not undergone any geometric transformation, for example by alternation of the values 0 and 1 representing respectively a normal or reversed order of the symbols, for each direction of the digital signal.
In a preferred implementation of the invention, which applies in particular to geometric manipulations of the images, for example, for printing them, the digital signal then being of dimension
2
, the geometric transformation applied is a transformation by axial or central symmetry of said symbols, a transformation by rotation through a multiple of 90° or a combination of said transformations.
In this same preferred embodiment of the invention, the transcoding method also comprises a step of transposition of a frequency sub-band having coefficients of low frequency in a first direction of the digital signal and high frequency in a second direction of the digital signal with a frequency sub-band with the same resolution level in the spectral breakdown, having coefficients of high frequency in said first direction and low frequency in said second direction, when the geometric transformation applied comprises a rotation through 90° or 270°.
This characteristic of
Berthelot Bertrand
Henry Felix
Majani Eric
Ahn Sam K.
Canon Kabushiki Kaisha
Chin Stephen
Fitzpatrick ,Cella, Harper & Scinto
LandOfFree
Geometric transcoding of a digital signal does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Geometric transcoding of a digital signal, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Geometric transcoding of a digital signal will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3318445