Abstract: The invention relates to a method for the processing of a multidimensional image which is formed by points representing objects of different intensity and are delimited from a background by way of their edges; this method includes steps for the segmentation (100; 30, 40) of the image into cells belonging to the background and cells belonging to the objects, for the application of criteria which impose that said cells (Ci) have substantially uniform intensities, that the cells neither intersect nor enclose any object edge, and that the object edges coincide with cells edges; this method includes a step for the presegmentation (30) of the image into a first number of cells while performing a classification of the image points in the various cells as a function of intensity gradients of these points, and an actual segmentation step (40) for reducing the first number of cells by classifying the cells in the form of pairs of adjoining cells (Cj, Ck), by evaluating a merger threshold (&lgr;0(j,k)) for the cells of

Abstract: The invention relates to an image processing method for determining a noise curve (NC) relating to an image, including acquisition (100) of an intensity image (J0) of pixels in which the noise is dependent on the signal, cutting (101) the intensity levels into substantially uniform (Mk) bands (QMk), determination (111) of statistical rules (f) which link distributions (Hk) of the standard noise deviations at the levels (QSk) to the intensities of said bands (QMk), determination (114), on the basis of statistical rules (f), of standard noise deviations (SBk) as a function of intensities (Mk), and determination (120) of the noise curve (NC) as a statistical rule (g) of the variations of the standard noise deviations (SBk) as a function of the intensity (Mk). The invention also relates to a method of reducing the noise which utilizes the standard noise deviation (SBk) and to a medical imaging apparatus which utilizes the noise reduction method.

Abstract: A digital image (J.sub.o) processing method for automatic extraction of strip-shaped objects, includes a skeletonization operation with steps for forming smoothed images (J.sub.i) at several scales (.sigma..sub.i), and, in each smoothed image (J.sub.i), extracting boundaries of objects, extracting potential median pixels (.OMEGA..sub.iP) associated with the center (.OMEGA..sub.i) of a circle of radius (R.sub.i) proportional (k) to the scale (.sigma..sub.i), tangent to boundaries at a pair of distinct pixels (E.sub.1, E.sub.2), and associated with a measure of dependability regarding alignment of the center (.OMEGA..sub.i) and pixels of the pair, extracting median pixels .OMEGA..sub.iM, and constructing skeletons of objects by tracking in the digital image (J.sub.MED) formed by the extracted median pixels. The step of extracting median pixels .OMEGA..sub.

Abstract: A method of processing a digital image representing ribbon-shaped objects of non-uniform intensity contrasting with a background of lower intensity includes an automatic segmentation phase having one or more morphological opening operations effected, respectively, with one or more three-dimensional structuring elements. The latter have a two-dimensional base parallel to the image plane and have a non-binary intensity function in a third dimension. Preferably, the automatic segmentation phase is carried out by means of a set of two-dimensional spatial structuring elements with a third intensity dimension. The set contains N anisotropic structuring elements oriented from .pi./N to .pi./N and one isotropic structuring element.

Abstract: A method for the temporal filtering of an image of a sequence of noisy digitized images includes the evaluation of an integration equation which produces the filtered intensity (P.sub.t.sup.C) of a given pixel ?A.sub.t (x,y)! in the present image (J.sub.t.sup.P) by evaluating the sum of the noisy intensity (I.sub.t.sup.P) of the present (t) and the product of a gain factor (K.sub.t.sup.P) and the difference of said noisy intensity of the present (I.sub.t.sup.P) and the causal filtered intensity (P.sub.t-1.sup.C). According to this evaluation, the gain factor (K.sub.t.sup.C) is a function of the causal gain factor (K.sub.t-1.sup.C) and of a continuity coefficient (.alpha..sub.t.sup.C) which measures the probability of intensity continuity between the present noisy intensity (I.sub.t.sup.P) and the causal filtered intensity (P.sub.t-1.sup.C). A device for carrying out this method includes a sub-assembly whose input (100) receives the present noisy sample (I.sub.t.sup.

Abstract: A method of processing digital angiographic images which enables automatic detection of stenoses includes a tracking step for the identification of pixels or points situated on the central lines of vessels, and a step for the detection of stenoses by measurement of the intensity of pixels along the central lines of the vessels. The stenoses detection step includes a sub-step for deciding whether a pixel zone is a potential stenosis zone when the intensity of its pixels exhibits a local variation so as to assume an intermediate value between the intensity of the pixels to both sides of this zone and the intensity of the pixels of the image background, a sub-step for defining an icon around potential stenosis zones, and a sub-step for the extraction of the pixels of edges of the vessel segments in the icons.

Abstract: An image processing method for automatic detection of regions of a predetermined type of cancer in an intensity image of a part of a human or animal body, includes, for a number of points of a part of said image, the determination of a set (referred to as a vector) of components formed by characteristic values derived from the intensity distribution around each point in said part of the image, and the use of a classification system for determining the probability of the point associated with said vector belonging to a region of said part of the image which corresponds to the predetermined type of cancer or to another region. An image processing device for carrying out such a method is utilized in conjunction with a digital medical imaging system.

Abstract: Digital image processing method for local determination of the center and the width of objects in the form of contrasting bands on a background. A digital image processing method, for representation of objects in the form of contrasting bands of substantially uniform intensity on a substantially uniform background, includes a step for the identification of pixels situated on the central lines of the objects, which step is referred to as a "tracking" step. This method is characterized in that the step includes a first filtering operation which is executed by applying, to each image, a series of N lozenge-type-bidimensional, selective, recursive low-pass filters having respective ones of the principal directions angulary spaced apart 180.degree./N in order to determine the direction of each band-shaped object segment as that in which the response of one of the filters is maximum.

Abstract: The invention relates to a method of learning which is carried out in a neural network operating on the basis of the gradient back-propagation algorithm. In order to determine the new synaptic coefficients with a minimum learning period, the invention introduces parameters which privilege corrections based on the sign of the error at the start of learning and which gradually induce less coarse corrections. This can be complemented by other parameters which favor a layer-wise strategy, accelerating the learning in the input layers with respect to the output layers. It is also possible to add a strategy which acts on the entire neural network.

Abstract: A method of processing images in order automatically to detect key pixels (K.sub.2, K.sub.3, K.sub.4) situated on the contour of an object (LV) in an initial image (I.sub.0), and a device for implementing this method in which there is storage, in the digitized initial image (I.sub.0), of the intensity of the pixels [A.sub.0 (x,y)] and of data of regions of the object (LV), classes (C.sub.1 C.sub.m); selection of pixels of interest (PI), on the contour, inside and outside the object (LV); generation of characteristics (E.sub.1 E.sub.k) for each of the pixels of interest (PI); classification of the pixels of interest (PI) into the classes (C.sub.1 C.sub.m); and selection of the key pixels (K.sub.2, K.sub.3, K.sub.4) from corresponding classes (C.sub.2, C.sub.3, C.sub.4).

Abstract: Unsupervised training method for a neural net and a neural net classifier device wherein test vectors are supplied to the neural net whose operational parameters are classified in a stochastic labeling procedure by mutually correlating the net's output activations for each test vector and on the basis thereof generating weighting factors that scale the probabilities when selecting a class at random. Once the test vectors are thus classified, the operational parameters of the net are modified in order to intensify the differences among the patterns of output activations for all test vectors.

Abstract: The invention relates to a method of manufacturing field effect transistors of gallium arsenide obtained by ion implantation of light donors, such as silicon or selenium, in a semi-insulating substrate of gallium arsenide. In order to reduce out-diffusion of the deep level (EL.sub.2) responsible for parasitic phenomena in the operation of the transistors, the method is characterized in that in addition oxygen ions are implanted in at least the region of the substrate intended to form the channel region of the field effect transistor. After implantation, the substrate is sintered at a temperature between 600.degree. and 900.degree. C. in either an enveloping substance or uncovered, and/or in an atmosphere of arsine.