Facsimile and static presentation processing – Facsimile – Specific signal processing circuitry
Patent
1984-08-15
1988-06-21
Groody, James J.
Facsimile and static presentation processing
Facsimile
Specific signal processing circuitry
358133, 358 13, H04N 704
Patent
active
047528273
DESCRIPTION:
BRIEF SUMMARY
The invention relates to a method or a device for the numerical coding of an image, more particularly a television image, made up by elementary points (pels or pixels) which are regularly distributed in the plane of the image and to which are allotted one or more radiometric qualities denoted as components C.sub.i.
The components C.sub.i can be, for example, the measurements of the radiometries of each pixel in N spectral channels supplied by suitable pickups staged over the spectrum of radio-electric frequencies; visible, infrared, X radiation, etc . . . They can also be quantities deduced from the measured radiometries, in a single spectrum zone of a predetermined spatial arrangement of pixels, by a linear transformation defining N independent channels.
The digitalization of an image for its processing or transmission causes a very large volume of information to appear which is always difficult to handle if acceptable fidelity is to be maintained.
In the example of colour television, it is considered that the density and precision of sampling for the practically distortion-free reproduction of a colour image requires about 6 million bits, at the rate of 16 bits per pixel. The numerical output of a television transmission on this basis implies a channel of 170M bits/second--i.e. the equivalent of 2,500 telephone calls. This is a ratio which is prohibitive in view of current cost levels and which would heavily penalise the economics of such a transmission system.
Attempts have therefore been made to evaluate numerical processing methods adapted to reduce such output and, for this purpose, to eliminate as far as possible the redundancy of the message. In practice this means avoiding the repetition of information already transmitted as long as the variations observed, in space or in the time between adjacent samples, are lower than certain values defined in relation to the required quality (signal-to-noise ratio, undesirable structures and fluctuations, sensitivity to errors, etc.)
Numerous proposals, using differential coding techniques, orthogonal transformations, statistical coding methods and a combination thereof have been tried out with differing degress of success, although none of them has been found completely satisfactory. It has also been proposed (G LOWITZ "Recognition of shapes and processing of images", AFCET-IRIA Congress, Feb. 21-23, 1978, pages 699-714) to adopt a method using a technique frequently known by the Anglo-Saxon term "clustering". In that method for the numeral coding of an image, more particularly a television image, windows are determined in the image which have a predetermined geometrical shape and comprise a certain number of elementary points; for each window regroupings are performed into classes of radiometries (spectral or spatial) of the ends of the vectors which represent the pixels or combinations of pixels of the image in space or N dimensions in accordance with a criterion of proximity; one defines for each class, from the individual values of the pixels forming the class, N typical radiometries which will be stored or transmitted to be used in the reconstruction of the image; to define each pixel a characteristic code of the class to which it belongs is also stored or transmitted enabling it to be given the typical corresponding radiometries for the reconstruction of the image.
It is a primary object of the invention to provide a method and device for numerical coding of images which are free from or at least less subject to the aforementioned disadvantages and which more particularly enable a high compression rate to be obtained.
To this end the invention proposes a method characterised in that a processing of high spatial frequencies is performed by a unit transformation one of whose coefficients, that of the new value corresponding to lower spatial frequencies, is associated with the components of different spectral channels for processing by a complementary classification.
The idea of proximity, the measurement of which is used as the criterion for regrouping th
REFERENCES:
patent: 4210927 (1980-07-01), Yumde et al.
patent: 4281312 (1981-07-01), Knudson
patent: 4302775 (1981-11-01), Widergren et al.
patent: 4558350 (1985-12-01), Murakami
G. E. Lowitz, "Compression des donnees images par reconnaissance des Formes et Clustering", Reconnaissance des formes et traitement des images, Congres AFCET-IRIA, Feb. 21-23, 1978, pp. 699-714.
F. J. In der Smitten, "Informations reduzierende Quellencodierung von Farbbildsignalen", Nachrichtentechnische Zeitschrift, No. 5, May 1974, pp. 176-181.
Cassagne Pierre
Lowitz Gabriel
Groody James J.
Harvey David E.
Matra, Societe Anonyme
LandOfFree
Method and device for the numerical coding of an image, more par 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 and device for the numerical coding of an image, more par, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and device for the numerical coding of an image, more par will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-932238