Jacques Duvernoy

Optical-digital processing of directional terrain textures invariant under translation, rotation, and change of scale.

Iso-energy directional signatures are measured in 2-D Fourier spectra of aerial photographs. They combine the respective advantages of both the ring-shaped and wedge-shaped features usually measured such spectra, namely, the sensitivity to the coarseness and directionality of textures. Different terrain textures characterize the spatial structures displayed in aerial photographs recorded in the visible domain. The study is restricted to three classes of textures associated with fields, woods, and urban regions. Invariant signatures are defined by an automated application of the technique of Fourier descriptors for plane closed curves to contour lines selected in the spectra of the photographs. In a second step the dimensionality of these descriptors is reduced by a partial expansion on an orthonormal polynomial basis. Their statistical significance is determined by using nonsupervised classification methods. As a result, an oblique basis of signatures matched to the three classes of textures is proposed, which is invariant under translation, rotation, and change of scale. The identification rate is optimized by a further orthogonalization of this basis, using the Gram-Schmidt procedure. A comparison between the results of terrain classifications performed using either the direct signatures or the two proposed bases is presented.

Stability and stationarity of cursive handwriting

Abstract Handwriting is considered as the output of a space variant imaging system equivalent to the writer. This system is described by the parameters of the synthesis of letters from a given model. A statistical analysis of the variations of this system on a set of successive pages leads to the extraction of writer features as well as the segmentation of the text into short-time stationarity domains to be related to “rhythms of writing”.

Effective optical processor for computing image moments at TV rate: use in handwriting recognition

An incoherent multiplexed optical processor is described that computes in real time the moments of images sensed by a TV camera and displayed on a monitor. The first ten moments are generated in parallel and are detected by a photodetector array. They are collected by a microcomputer for subsequent pattern recognition use. Calibration and resolution problems are discussed with regard to the effective 2% system accuracy. Its use in real-time processing of one hundred handwritten pages shows the processor to be a practical moment generator.

Processing measurements of the directional content of Fourier spectra.

Directional analysis and filtering make use of the most straightforward properties of optical Fourier transforms. A quantitative study of the detectability of angular maximums in Fourier spectra and its relationships with the shape and distribution of elements in the object is proposed. The problem is modeled by using a simple object which consists of the superimposition of two random distributions of rectangular grains, one being rotated with respect to the other. Computing the Fourier spectrum of this object allows the expression of the amount of light integrated through a wedge which scans the spectrum. An index of angular separability of the two distributions is made up from such measurements. It is shown to depend on the rotation angle as well as the grain shape. Using an additional spatial filter can improve this index. The influence of its radius is studied. Experimental results obtained either in real time with optical fibers or on photographic records with a microdensitometer are compared with the theoretical ones. They show the necessity of the implementation of optimal estimation schemes to reduce the influence of the noise. Two linear least squares filters are used: an angular Wiener filter and an autoregressive one. In the case of a high signal-to-noise ratio, the Wiener filter reduces to a Laplacian filter which does not depend on the shape of the grains.

RMN 13C: modèle structural et statistique de décomposition des effets ortho en series benzènique et aromatiques polycycliques monosubstituées

Abstract From long range effects, detected by NMR 13 C in particular monosubstituted benzenes XC 6 H 5 of the general form X  YC 6 H 4 G, a new approach is proposed for the ortho effect. Neutral and charged (positive or negative) substituents X are used to define this model. For the ortho position, the substituent chemical shift (SCS) is decomposed into two parts: a first term proportional to the para SCS, and a second term, W X , accounting for proximity effects. To calculate the different parameters, the X groups are separated into series characterized by a common key-atom. The generality of the term W X is demonstrated with other aromatic families, such as 2X-naphthalenes, 2X-anthraquinones and benzothiazoles, proving the validity of the proposed model. This model is also applied to the carbon C β of monosubstituted ethylenes.

Bistabilité, multistabilité et chaos en longueur d’onde

AnalyseDes études d’instabilités et de chaos en optique se sont développées, liées à la mise en œuvre de sources de fonctions non linéaires et de circuiterie de rétroaction optique. Des phénomènes liés aux variations d’énergie dans un faisceau monochromatique ont été étudiés, faisant appel à des non-linéarités intrinsèques ou à des dispositifs hybrides. Nos travaux concernent les phénomènes chaotiques observables sur la variable longueur d’onde. Celleci offre des possibilités nouvelles par la maîtrise des non-linéarités en longueur d’onde et la dynamique d’excursion des signaux. Des dispositifs bi- et multistables en longueur d’onde sont présentés ainsi qu’une étude des propriétés des différents types de chaos qu’ils peuvent générer. Les perspectives qui sont discutées en conclusion concernent la mise en œuvre de la variable longueur d’onde pour un simulateur de chaos associé à d’autres grandeurs physiques, le codage multivalué de signaux en longueur d’onde ainsi que les formes ultérieures de chaos résultant d’un retard (optique) dépendant du signal.AbstractUnstable and chaotic phenomena have lead to a number of studies in the area of optics which allows easy experimental verifications owing to the use of optical sources and feedback, and optically-induced non linearities. Most of those studies have dealt with hybrid or all-optical systems working in monochromatic light and whose dynamic variable of concern was the energy. The work presented here deals with chaotic phenomena linked to the wavelength variable. The latter opens up new possibilities inherent to the high flexibility in inducing wavelength non linearities and their control, and to the greater dynamic excursion of the signals. Wavelength bistable and multistable devices are presented, together with a discussion of the various routes to chaos they can generate. The prospects given in the conclusion deal with the use of the wavelength variable in a chaos generator, the multivalued coding in wavelength of signals, and the chaos that would be generated by a system with a feedback featuring a signal-dependent optical delay.

Optical statistical classifiers using coherent and incoherent light.

Breast thermograms are digitally classified by means of a principal component analysis of a training set of breast spectral differences obtained in coherent light. The results are compared to the usual partition in categories of medical diagnosis. The statistical operators initiating the classification being expressed in terms of spatial frequencies, the thermograms can be optically classified by using optical representations of the operators that play the role of filters. In coherent light one deals with a conventional spatial frequency filtering, and in incoherent light the thermograms are processed by imaging through suitable coding pupils. In both cases the principal components are easily computed from the measurement of luminous intensities. Such analog computers are expected to increase the throughput of processed images by avoiding needless digitization.

Statistical Matching of the Directional Content of an Illumination Source Application to the Restoration of the Legibility of Roman Inscriptions

The visibility of erased inscriptions on Roman stones is improved by illuminating them with a lateral source of light. The optimal angular location and incidence of the source depend on the directional content of the graven letters as well as their degree of erosion. A general multidirectional illumination scheme has been worked out by statistically matching the respective intensities of a set of n primary sources surrounding the stone. The amount of directional information accounted for by each primary source with respect to the corresponding primary image of the stone, disclosed by the source. Multispectral information compression techniques can therefore be applied to the n-dimensional space spanned by the general source. Principal images are computed from the analysis of variance of the directional information in this space. Global or local criteria have to be implemented, depending on degree of erosion of the inscriptions. Full resolution principal images are synthetized by linear combination of photographs of the primary images. Application to the deciphering and reading of Raman inscriptions are presented. The principle of a real time scanner based on the multiplexing properties of white light is introduced as a conclusion. It takes advantage of the achromaticity of the studied objects for coding a spatial frequency scanner upon the chromatic variable. The princi-ple of a generalized source, that illuminates a scene in a pattern recognition context, results from mixing two features : the directional content is statiscally matched to that of the scene, and the chromatic content allows a spatial frequencies sampling

Processing optical data matrices.

The cross-correlation matrix of a set of signals is obtained by multiple Fourier holography. Various sources of error make this kind of matrix not directly interpretable. Global methods that correct deterministic fluctuations as well as random perturbations are proposed. They do not take into account any individual element of the matrix, but they minimize the over-all variance. The efficiency of these methods is illustrated by the example of the classification of sonagrams.