Pulse or digital communications – Repeaters – Testing
Reexamination Certificate
1998-09-01
2001-11-13
Britton, Howard (Department: 2613)
Pulse or digital communications
Repeaters
Testing
C382S248000
Reexamination Certificate
active
06317451
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the invention
The present invention concerns a method of quantizing a multidimensional sampled signal comprising at least one multidimensional signal of given dimension N, the or each sample including N components, the method including the steps of:
a) for each sample, converting each component into a set of transformed coefficients in N known numbering bases;
b) combining the sets of transformed coefficients; and
c) progressively quantizing the combination of the sets of transformed coefficients.
The method is intended in particular for compressing signals representing color pictures, in particular TV pictures, stereophonic sound or data resulting from transforms such as Fourier transforms, discrete cosine transforms and wavelet transforms.
2. Description of the Prior Art
In the present context, each signal is formed of a set of transformed or not-transformed multidimensionsal samples of given dimension N. Thus each sample includes N components.
In the case of compressing a set of multidimensional sampled signals, existing compression schemes treat each component of the signal as an independent sub-signal. Each sub-signal is first transformed and then quantized independently of the other transformed sub-signals of the same multidimensional sampled signal.
For example, the three color planes (red, green, blue) of a color television picture are usually quantized independently of each other.
A quantizing method is described in “Embedded Image Coding Using Zerotrees of Wavelet Coefficients” by J. Shapiro (IEEE Trans. on Signal Processing, vol. 41, No 12, December 1993).
The work of J. Shapiro concerns a method of integrated progressive quantization of images with a plurality of levels of gray that have been subjected to wavelet transformation.
The proposed quantization process allows progressive transmission of the compressed data. Progressive quantization means that degraded information can be partially re-established from the compressed bit frame of which only a part has been received.
Accordingly, with a progressive quantizing process, the frame of bits obtained at a high compression level is the start of the frame of bits obtained at a lower compression level. In other words, increasing the compression level amounts to truncating the bit frame so as to retain only its beginning.
The algorithm of J. Shapiro transposed for processing color images, consisting in processing each color successively, does not allow for the correlations between the colors and this degrades compression performance.
A solution to these problems proposed by A. Said and W. A. Pearlman in “A New Fast and Efficient Image Codec Based on Set Partitioning In Hierachical Trees” (IEEE Trans. on Circuits & Systems for Video Technology, vol. 6, No 3, June 1996) aims among other things to transform characteristic variables corresponding to each of the colors beforehand by decorrelating them, in particular using a main component analysis method. Using this method, the three colors are transformed beforehand into a luminance component and two chrominance components. A. Said and W. A. Pearlman propose to apply a Karuhnen-Lowe transform to the various bands to decorrelate them.
However, using decorrelation necessitates an additional transformation, which is relatively time-consuming. Moreover, the operation is not reversible because of rounding off problems, and this degrades the quality of the reconstituted image, preventing its use in some fields, such as the medical field, for statutory reasons.
Using the above methods, the data frames include information relating to each of the colors in succession or, if the signal has been transformed beforehand, to luminance and then chrominance. Accordingly, the frame must be transmitted completely to enable satisfactory reconstitution of the image.
Any partial transmission of only a truncated frame yields only first information which on its own is insufficient to reconstitute a correct image. There is then the risk of losing all the information for one color plane or, if the signal has been transformed beforehand, some chrominance information.
To solve this problem A. Said and W. A. Pearlman propose mixing the components formed from the transformed characteristic variables by interleaving them following a particular form of sorting before quantizing the components combined in this way.
However, this method is relatively time-consuming because it necessitates the use of a particular sorting process.
Further, despite the interleaving of the components, quantizing being effected on each scalar coefficient transformed independently of the component to which it corresponds, a truncated data frame does not always convey the same number of information items for all the components representative of the image.
Accordingly, the above methods are not completely satisfactory.
An aim of the invention is to propose a method of quantizing a multidimensional sampled signal allowing for the high level of interdependence of the components of the multidimensional samples combined with effective exploitation of a truncated frame of the quantized signal.
SUMMARY OF THE INVENTION
To this end, the invention consists in a method of quantizing a multidimensional sampled signal including at least one multidimensional sample of given dimension N, the sample or each sample including N components, the process including the steps of:
a) for each sample, transforming each component into a set of transformed coefficients in N known numbering bases;
b) combining the sets of transformed coefficients; and
c) progressively quantizing the combination of the sets of transformed coefficients; wherein
for combining the sets of transformed coefficients it includes the following two successive steps:
A) creating, for the multidimensional sample or for each multidimensional sample, a vector the coefficients of which are N-tuplets formed by the transformed coefficients of the same rank i in the N bases of transformation of the characteristic variables; and
B) combining the vectors of N-tuplets;
the method finally including the progressive quantizing of the combination of the vectors of N-tuplets, progressive quantizing being effected on the N-tuplets.
In particular embodiments, the method may have one or more of the following features:
the transformation of each component of each sample into a set of transformed coefficients in N known numbering bases is effected by direct transformation of each sample in a numbering base the base vectors of which are N-tuplets;
the vectors of N-tuplets are combined in the order of the multidimensional samples in the signal;
the progressive quantization step includes a step of conditional coding of certain elements of the same N-tuplet in accordance with the other elements of the N-tuplet concerned;
the multidimensional samples are obtained by preliminary transformation of a set of multidimensional information items each including N characteristic variables, the preliminary transformation effected on the multidirectional information items consisting of a set of independent transforms each operated on one of N planes consisting of characteristic variables of the same rank of the multidimensional information; and
the signal is a signal representative of a digital color image.
The invention will be better understood after reading the following description given by way of example only and with reference to the drawings.
REFERENCES:
patent: 5422736 (1995-06-01), Katayama
patent: 5883978 (1999-03-01), Ono
patent: 5886741 (1999-03-01), Srivastava
patent: 6018366 (2000-01-01), Asai et al.
patent: 0 107 072 A2 (1984-05-01), None
J. M. Shapiro, “Embedded Image Coding using Zerotrees of Wavelet Coefficients”, Dec. 1, 1993, IEEE Transactions on Signal Processing, vol. 41, Nr. 12, pp. 3445-3462.
Alcatel
Britton Howard
Sughrue Mion Zinn Macpeak & Seas, PLLC
LandOfFree
Method of compressing a multidimensional sampled 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 Method of compressing a multidimensional sampled signal, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method of compressing a multidimensional sampled signal will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2574817