<html>  <head> <meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1"> <meta name="Template" content="P:\OFFICE97\OFFICE\html.dot"> <meta name="GENERATOR" content="Microsoft FrontPage 3.0"> <title>PUBLICATIONS-du projet VISIVOL-CAPORAL-GIQUAL</title> </head>  <body bgcolor="#FFFFFF" LINK="#802809" VLINK="#0a6622" ALINK="#a2bdfc"> <p align="center"><font size="5" face="Arial"><b>VISIVOL</b></font></p>  <table border="0" cellpadding="4" cellspacing="4" width="99%" height="7812">   <tr>     <td colspan="3" width="15%" bgcolor="#FCF6CF" style="border-right: medium solid rgb(252,246,207)" height="20"><a name="BM2"></a><a href="pubvisivol.htm#BM3"><font size="2">article        suivant</font></a></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="31"><font size="1" face="Arial"><b>Type de document</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="31"><font size="2">Article - Revue     scientifique  comit de lecture </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="32"><font size="1" face="Arial"><b>Titre</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="32"><font size="2">Caractrisation en     ligne de la granulomtrie de produits de broyage de bl par l'utilisation d'attributs     globaux issus de l'analyse d'images </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="16"><font size="1" face="Arial"><b>Title</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="16"><font size="2">Particle size     characterization of in-flow milling products by video image analysis using global features     </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Auteur</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">NOVALES B./ GUILLAUME     S./ DEVAUX M.F./ CHAURAND M. </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="31"><font size="1" face="Arial"><b>Anne de publication</b></font><font size="2">     </font></td>     <td valign="top" colspan="2" width="90%" height="31"><font size="2">1998 </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Source</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">J.Sci.Food Agric., 78 </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Langue</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">ENG </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Page</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">p. 187-195 </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="152"><font size="1" face="Arial"><b>Rsum</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="152"><font size="2">L'objectif de ce     travail tait de caractriser la granulomtrie de bl en cours de broyage par analyse     d'image. Quatre classes de produits ont t obtenues en faisant varier l'cartement des     rouleaux du deuxime broyeur du moulin. Le dispositif d'analyse d'image a t install     en ligne et 1300 images ont t enregistres pour chaque classe. Trois mthodes     globales d'analyse d'image ont t testes : l'ouverture morphologique, les longueurs     de lignes  niveaux de gris constant, l'interdpendance spatiale des niveaux de gris.     Une analyse discriminante a t applique sur les donnes extraites par ces trois     mthodes. Plus de 77% des chantillons, de l'ensemble d'apprentissage ou de l'ensemble     de validation, ont t correctement classs. Les meilleurs rsultats ont t obtenus     par le traitement des donnes issues de par la distribution par ouverture d'une part, et     des matrices de co-occurrences d'autre part. Les performances en classification atteignent     81% avec seulement 3 paramtres pour la distribution par ouvertures et, 83% avec 3     paramtres des matrices de co-occurrences. </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="152"><font size="1" face="Arial"><b>Abstract</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="152"><font size="2">The objective of this     work was to characterise the particle size of milling products by image analysis. Four     classes of milling products were obtained by varying the roll gap of the second break roll     of the mill. Images were acquired by using an in-flow imaging system implemented in the     mill, and 1300 images were recorded for each class. Three methods of image analysis were     investigated: morphological opening, constant grey level run lengths and grey level     spatial interdependences. Discriminant anlyses were applied to the data extracted from the     images by the three methods in order to identify each class of milling product. More than     77% of the samples were correctly assigned to their group both for the calibration and     validation sets. The best results were obtained by applyng morphological openings or by     computing parameters from the co-occurrence matrices. The number of correct     classifications rose to 81% of samples with only three variables selected for the opening     curves and to 83% with three co-occurrence parameters. </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Mots cls</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">COOCCURENCE/GRANULOMETRIE     / BLE / TRAITEMENT D'IMAGES/ </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Diffusion</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">Diffusion tous publics     </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Cote</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">98/0661 </font></td>   </tr>   <tr>     <td width="15%" style="border-right: medium solid rgb(252,246,207)" height="21"></td>     <td colspan="2" height="21"></td>   </tr>   <tr>     <td width="15%" bgcolor="#FCF6CF" style="border-right: medium solid rgb(252,246,207)" height="35"><a name="BM3"></a><a href="pubvisivol.htm#BM2"><font size="2">article        prcdent</font></a></td>     <td bgcolor="#FCF6CF" height="35"><a href="pubvisivol.htm#BM4"><font size="2">article        suivant</font></a></td>     <td bgcolor="#FCF6CF" height="35"><font size="2"><a href="pubvisivol.htm#BM1">dbut        de la page</a> </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="31"><font size="1" face="Arial"><b>Type de document</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="31"><font size="2">Article - Revue     scientifique  comit de lecture </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Titre</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">Classifictaion de     produits ganulaires utilisant la fusion de niveau haut de paramtres d'images </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Title</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">Classification of     granular product using high level fusion of vision features </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Auteur</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">ROS F./GUILLAUME     S./BELLON MAUREL V. </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="31"><font size="1" face="Arial"><b>Anne de publication</b></font><font size="2">     </font></td>     <td valign="top" colspan="2" width="90%" height="31"><font size="2">1997 </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Source</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">Journal of     agricultural engineering research, vol. 68 </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Langue</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">ENG </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Page</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">p. 115-124 </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="167"><font size="1" face="Arial"><b>Rsum</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="167"><font size="2">La caractrisation     de produits granulaires par analyse d'image est un problme difficile  cause de la     diversit des donnes  extraire. Ce problme de classification peut tre rsolu par     deux approches. Une de ces approches consiste  agrger l'information qualitative     contenue dans chaque paramtre, en la considrant comme un capteur virtuel. C'est une     approche en trois tapes : d'abord, pour chaque paramtre (c'est--dire capteur     virtuel), les chantillons reoivent une probabilit d'appartenir  une classe     (clustering) ; puis ces probabilits sont agrges pour donner  chaque chantillon     la probabilit d'appartenir  chaque classe (rseau de neurone supervis) ; enfin,     l'chantillon est affect  la classe qui a la probabilit maximale. Cette procdure     est applique pour classer des semoules obtenues en broyant du bl dur. Trois classes     sont dfinies : 300, 400 et 500 m. Le taux de classification correcte est de 80%. Cette     mthodologie est particulirement intressante car elle donne un rsultat satisfaisant     et est totalement adaptable : l'ajout de nouveaux paramtres au procd de     classification ne demande le renouvellement que d'une partie de la procdure. </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="182"><font size="1" face="Arial"><b>Abstract</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="182"><font size="2">The characterization     of a granular product based on image analysis can be a difficult problem because it often     requires the combination of a large number of features of different natures extracted from     the image. This problem of classification can be solved by two approaches. One of these     approaches consists of aggregating the qualitative information which is obtained by     considering each individual feature, as a virtual sensor. This is a triple-step system :     first, for each feature (i.e. virtual sensor), the samples are given a probability of     belonging to a class (clustering) ; second, these probabilities are aggregated in order to     give a global probability of the sample of belonging to each class (supervised neural     network) ; third, the sample is assigned to the class which shows the maximal global     probability. This procedure was applied to classify semolina samples. These were obtained     by grinding wheat grains. Three classes were defined using three grinding roll gaps of     0.3, 0.4 and 0.5 mm, respectively. The average of correct classification was better than     80 %. This methodology is particularly interesting because it gives a very satisfactory     result and is quite versatile: new features added to the classification process require an     update of one part of the procedure only. </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Mots cls</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">/VISION / CAPTEUR /     TRAITEMENT D'IMAGES / AIDE A LA DECISION / IAA </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Diffusion</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">Diffusion tous publics     </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Cote</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">97/0672 </font></td>   </tr>   <tr>     <td width="15%" style="border-right: medium solid rgb(252,246,207)" height="21"></td>     <td height="21"></td>     <td height="21"></td>   </tr>   <tr>     <td width="15%" bgcolor="#FCF6CF" style="border-right: medium solid rgb(252,246,207)" height="35"><a name="BM4"></a><a href="pubvisivol.htm#BM3"><font size="2">article        prcdent</font></a></td>     <td bgcolor="#FCF6CF" height="35"><a href="pubvisivol.htm#BM5"><font size="2">article        suivant</font></a></td>     <td bgcolor="#FCF6CF" height="35"><font size="2"><a href="pubvisivol.htm#BM1">dbut        de la page</a> </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="31"><font size="1" face="Arial"><b>Type de document</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="31"><font size="2">Article - Revue     scientifique  comit de lecture </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Titre</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">Combiner des     caractristiques individuelles et globales pour caractriser de populations de produits     granulaires </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Title</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">Combining global and     individual image features to characterize granular product populations </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Auteur</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">ROS F./GUILLAUME     S./RABATEL G./SEVILA F./BERTRAND D. </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="31"><font size="1" face="Arial"><b>Anne de publication</b></font><font size="2">     </font></td>     <td valign="top" colspan="2" width="90%" height="31"><font size="2">1997 </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Source</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">Journal of     chemometrics, vol. 11 </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Langue</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">ENG </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Page</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">p. 483-500 </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="152"><font size="1" face="Arial"><b>Rsum</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="152"><font size="2">La caractrisation     de populations de produits granulaires par analyse d'image est un problme dlicat car     il ncessite l'extraction et la combinaison de diffrents types de caractristiques.     Nous proposons de l'tudier dans un cadre gnral en considrant les tapes de     l'analyse d'image, du traitement de l'information extraite et de la prise de dcision     pour effectuer la classification. Dans cet article nous traitons particulirement du     systme de dcision. Celui-ci est bas sur une approche hirarchique : un tage     gnraliste est conu pour donner un sous ensemble approximatif de solutions possibles,     puis des systmes spcialiss permettent de choisir parmi ces solutions  quelle     classe sera affect l'chantillon. Les descripteurs de l'chantillon  classer sont     diffrents pour chacun de ces systmes, ils sont choisis au cours d'une procdure de     paramtrage en fonction de la classification  raliser. Cette mthode a t     applique sur un problme de classification d'chantillons de farine en cours de     broyage. Trois types d'outils de classification ont t compars : l'analyse     discriminante, les k plus proches voisins, les rseaux de neurones. </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="182"><font size="1" face="Arial"><b>Abstract</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="182"><font size="2">The characterisation     of granular product populations using image analysis is a difficult problem because it     often requires the extraction and the combination of many different features. We propose     to study in a general way these problems of granular product classification, considering     the image analysis phase, the processing of the information extracted, and the     decision-making. In this paper, we focus rather on the decision system development. It is     based on a hierarchical approach of the problem, including a generalist system whose     outputs are ambiguous (an approximative solution), connected to specialist systems trained     to give non-ambiguous solutions. The inputs of the generalist system are the components of     a vector containing the most important information for discriminating all the decision     classes, while the inputs of the specialist systems are those which best distinguish a     given class from another. This strategy enables to overcome the multi-class aspect of the     problem. It is independent of the choice of the techniques to select the pertinent     information and to take the decision. This method is applied in the framework of a meal     classification where three types of classifier (discriminant analysis, k nearest     neighbours and multilayer neural networks) are compared. </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="32"><font size="1" face="Arial"><b>Mots cls</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="32"><font size="2">PRODUIT     GRANULAIRE/TRAITEMENT D'IMAGES / CLASSIFICATION / SELECTION / TRAITEMENT DE L'INFORMATION     / ANALYSE DISCRIMINANTE / RESEAU DE NEURONES/ </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Diffusion</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">Diffusion tous publics     </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Cote</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">97/0792 </font></td>   </tr>   <tr>     <td width="15%" style="border-right: medium solid rgb(252,246,207)" height="21"></td>     <td height="21"></td>     <td height="21"></td>   </tr>   <tr>     <td width="15%" bgcolor="#FCF6CF" style="border-right: medium solid rgb(252,246,207)" height="35"><a name="BM5"></a><a href="pubvisivol.htm#BM4"><font size="2">article        prcdent</font></a></td>     <td bgcolor="#FCF6CF" height="35"><a href="pubvisivol.htm#BM6"><font size="2">article        suivant</font></a></td>     <td bgcolor="#FCF6CF" height="35"><a href="pubvisivol.htm#BM1"><font size="2">dbut        de la page</font></a></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="31"><font size="1" face="Arial"><b>Type de document</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="31"><font size="2">Article - Revue     scientifique  comit de lecture </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="32"><font size="1" face="Arial"><b>Titre</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="32"><font size="2">Extraction et     pr-traitement de donnes images  partir d'indices de reprsentativit et de     classification, appliques  la caractrisation de produits granulaires </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="32"><font size="1" face="Arial"><b>Title</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="32"><font size="2">Building and     pre-processing of image data using indices of representativeness and classification,     applied to granular product characterization </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Auteur</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">ROS F./GUILLAUME     S./BELLON MAUREL V./BERTRAND D. </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="31"><font size="1" face="Arial"><b>Anne de publication</b></font><font size="2">     </font></td>     <td valign="top" colspan="2" width="90%" height="31"><font size="2">1997 </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Source</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">Journal of     chemometrics, vol. 11 </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Langue</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">ENG </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Page</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">p. 469-482 </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="212"><font size="1" face="Arial"><b>Rsum</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="212"><font size="2">La caractrisation     de produits granulaires par analyse d'image est un problme difficile, car la taille d'un     chantillon est difficile  dfinir (doit-on utiliser le poids ou le nombre de     particules ? ) et  cause de la diversit des donnes  extraire. Une dmarche en     trois tapes est utilise : extraction de donnes, pr-traitement et classification     des chantillons. Dans cet article, nous nous intressons  la deuxime tape, une     fois que l'image a t extraite et transforme en histogrammes avec un grand nombre     d'intervalles. La mthode propose permet de construire des chantillons de taille     optimale et de crer des vecteurs de donnes appropries pour la troisime tape.     L'originalit de la mthode rside dans la supervision de l'tape de traitement avec     prise en compte de l'objectif final, c'est--dire la discrimination entre classes. Des     indices de stabilit et de discrimination sont crs pour construire de nouveaux     histogrammes. Pour dterminer la taille optimale, des indices de reprsentativit et de     pouvoir de classifications sont utiliss. Cette procdure a t teste sur des images     de produits craliers diviss en trois classes. La taille optimale de l'chantillon,     dtermine par l'indice de reprsentativit est de 18 images, et tombe  13 images en     utilisant l'indice de classification. Dans cet exemple, les paramtres, considrs     indpendamment, ne sont pas assez pertinents pour rpondre au problme (la meilleure     classification est correcte  60%). Il est ncessaire de dvelopper une stratgie pour     combiner ces paramtres. Elle sera prsente dans un autre document. </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="197"><font size="1" face="Arial"><b>Abstract</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="197"><font size="2">The characterization     of granular products using image analysis is complex, as defining sample size is a very     difficult task (should one use weight or number of particles ?) and because of the     diversity of the date which can be extracted from the image. A three-step procedure is     applied : data extraction, data preprocessing and sample classification. We deal with the     second step, once the image data have been extracted and gathered into histograms with a     large number of intervals. The method we propose allows both the building of optimal size     samples and the creation of data vectors appropriate for the third step.The originality of     the method lies in the supervision of the data processing by taking into account the final     goal, the discrimination into classes. Indices of stability and discrimination are created     to build new histograms. To determine the optimal sample size, indices of     representativeness and classification are used. This process has been tested on mill     product images which are divided into three classes. The optimal sample size given by the     representativeness index is 18 images, whereas it drops to 13 using the clasification     index. For this example the features, if considered independently, are not informative     enough to solve the problem (the best classification performance is 60 %). It is necessary     to develop a strategy where features are combined. This strategy is presented in a     separate paper. </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Mots cls</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">/CAPTEUR / TRAITEMENT     D'IMAGES / IAA </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Diffusion</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">Diffusion tous publics     </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Cote</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">97/0673 </font></td>   </tr>   <tr>     <td width="15%" style="border-right: medium solid rgb(252,246,207)" height="21"></td>     <td height="21"></td>     <td height="21"></td>   </tr>   <tr>     <td width="15%" bgcolor="#FCF6CF" style="border-right: medium solid rgb(252,246,207)" height="35"><a name="BM6"></a><a href="pubvisivol.htm#BM5"><font size="2">article        prcdent</font></a></td>     <td bgcolor="#FCF6CF" height="35"><a href="pubvisivol.htm#BM7"><font size="2">article        suivant</font></a></td>     <td bgcolor="#FCF6CF" height="35"><a href="pubvisivol.htm#BM1"><font size="2">dbut        de la page</font></a></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="31"><font size="1" face="Arial"><b>Type de document</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="31"><font size="2">Article - Revue     scientifique  comit de lecture </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Titre</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">Caractrisation de     produits de mouture par analyse d'image en ligne </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Title</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">Characterisation of     mill products by analysis of in flow digitalized images </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Auteur</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">GUILLAUME S./ROS     F./CHAURAND M./BELLON MAUREL V./ABECASSIS J. </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="31"><font size="1" face="Arial"><b>Anne de publication</b></font><font size="2">     </font></td>     <td valign="top" colspan="2" width="90%" height="31"><font size="2">1996 </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Source</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">Journal of food     engineering, n 27 </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Langue</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">ENG </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Page</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">p. 311-322 </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="107"><font size="1" face="Arial"><b>Rsum</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="107"><font size="2">Un capteur pour     caractriser les produits granulaires dans les industries agro-alimentaires a t     conu et test dans une semoulerie pilote. Il est compos de : un systme mcanique     qui prlve une part reprsentative du flux ; une camra CCD qui assure la prise     d'images. Un ensemble de logiciels d'analyse d'images et de traitement des donnes. La     mthode consiste  comparer l'chantillon avec une classe de qualit. Le systme de     dcision est construit par un apprentissage : ce sont les cas rels prsents au     systme qui permettent de le configurer. Trois classes de qualit ont t dfinies,     elles correspondent  l'cartement des rouleaux (0.30, 0.40, et 0.50) du broyeur de     tte de la semoulerie. Dans ces conditions les rsultats en classification sont     suprieures  80 %. </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="107"><font size="1" face="Arial"><b>Abstract</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="107"><font size="2">A sensor has been     designed and tested in a pilot mill to characterize granular products in the food     industry. It consists of : a mechanical system which takes a representative part of the     product, a CCD camera to capture images, a software package for image analysis and data     processing. The method consists of comparing a sample with a predeterminded quality     class'. The decision system is built on example learning : real cases fed into the system     allow its configuration. Three quality classes have been defined, they correspond to the     rolls gap (0.30, 0.40, 0.50 mm) of the first break rolls of a semolina pilot mill. In     these conditions, the classification accuracy rate achieved by the system is higher than     80 %. </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Mots cls</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">PRODUIT GRANULAIRE/     SEMOULERIE/IAA / CAPTEUR / CAMERA / CONTROLE DE QUALITE/ </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Diffusion</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">Diffusion tous publics     </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Cote</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">96/0012 </font></td>   </tr>   <tr>     <td width="15%" style="border-right: medium solid rgb(252,246,207)" height="21"></td>     <td height="21"></td>     <td height="21"></td>   </tr>   <tr>     <td width="15%" bgcolor="#FCF6CF" style="border-right: medium solid rgb(252,246,207)" height="35"><a name="BM7"></a><a href="pubvisivol.htm#BM6"><font size="2">article        prcdent</font></a></td>     <td bgcolor="#FCF6CF" height="35"><a href="pubvisivol.htm#BM8"><font size="2">article        suivant</font></a></td>     <td bgcolor="#FCF6CF" height="35"><a href="pubvisivol.htm#BM1"><font size="2">dbut        de la page</font></a></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="31"><font size="1" face="Arial"><b>Type de document</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="31"><font size="2">Communication  un     congrs </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Titre</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">Caractrisation de     produits granulaires : l'apport de l'analyse d'image pour la mouture du bl </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Title</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">Characterization of     granular products: the benefit of image analysis for wheat maslin </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Congrs</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">Procds de broyage,     Toulouse, 14-15 fvrier 1996 </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Auteur</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">GUILLAUME S./NOVALES     B./ABECASSIS J./DEVAUX M.F. </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="31"><font size="1" face="Arial"><b>Anne de publication</b></font><font size="2">     </font></td>     <td valign="top" colspan="2" width="90%" height="31"><font size="2">1996 </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Editeur</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">Groupe franais de     gnie des procds, Nancy </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Langue</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">FRE </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Page</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">p. 81-86 </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="107"><font size="1" face="Arial"><b>Rsum</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="107"><font size="2">Dans les semouleries,     la fragmentation des grains de bl dur est ralis par des appareils  cylindres. Le     rglage de ces machines conditionne le rendement pondral en semoule et est donc d'une     importance capitale d'un point de vue conomique. La vision artificielle est un capteur     permettant la caractrisation granulomtrique des produits de broyage de forme     quelconque. Pour viter toute prparation de l'chantillon, les images ont t     acquises en ligne, en flux tombant au rythme du procd. Quatre classes dfinies par     l'cartement des rouleaux du broyeur de tte (0.35, 0.45 et 0.50 mm) ont t     caractrises par leur distributions granulomtriques. Le traitement des donnes     extraites des images permet de prdire les classes avec un taux de succs suprieur      80 %. </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="107"><font size="1" face="Arial"><b>Abstract</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="107"><font size="2">In cereal milling     processes, the breaking of hard wheat grains for the production of semolina is carried out     by break rolls. The milling yield depends on the setting of the rolls and is of major     importance from an economical point of view. Artificial vision is a sensor allowing the     granulometric characterisation of milling products of various particles shapes. Images of     the milling products in flow rate were captured on-line in order to avoid any sample     preparation. Four classes defined by the gaps of the first break rolls (0.35, 0.45 and     0.50 mm) were characterised by their granulometric distributions. The images data     processing and the classification procedure made it possible to correctly identify more     than 80 % of the samples. </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Mots cls</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">PRODUIT GRANULAIRE/     ANALYSE D'IMAGE EN LIGNE/TRAITEMENT D'IMAGES / CLASSIFICATION/ </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Diffusion</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">Diffusion tous publics     </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Cote</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">96/0087 </font></td>   </tr>   <tr>     <td width="15%" style="border-right: medium solid rgb(252,246,207)" height="21"></td>     <td height="21"></td>     <td height="21"></td>   </tr>   <tr>     <td width="15%" bgcolor="#FCF6CF" style="border-right: medium solid rgb(252,246,207)" height="35"><a name="BM8"></a><a href="pubvisivol.htm#BM7"><font size="2">article        prcdent</font></a></td>     <td bgcolor="#FCF6CF" height="35"><a href="pubvisivol.htm#BM9"><font size="2">article        suivant</font></a></td>     <td bgcolor="#FCF6CF" height="35"><a href="pubvisivol.htm#BM1"><font size="2">dbut        de la page</font></a></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="31"><font size="1" face="Arial"><b>Type de document</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="31"><font size="2">Thse de doctorat </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Titre</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">Mthodes pour la     caractrisation de produits granulaires  partir de l'analyse d'image </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Title</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">A guide for the     conception of classification systems of granular product populations using image analysis </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Auteur</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">ROS F. </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="31"><font size="1" face="Arial"><b>Anne de publication</b></font><font size="2">     </font></td>     <td valign="top" colspan="2" width="90%" height="31"><font size="2">1995 </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Diplme</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">Doctorat Gnie des     procds bio-industriels, ENGREF Paris </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Langue</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">FRE </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Page</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">219 p. </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="197"><font size="1" face="Arial"><b>Rsum</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="197"><font size="2">Un guide pour la     conception de systmes de classification de populations granulaires  partir de     l'analyse d'image a t ralis. Il permet  un utilisateur non averti de concevoir     des capteurs intelligents capables d'apprcier des produits granulaires de manire     similaire  une expertise. Relier des images de produits granulaires  une dcision     qualitative demande des considrations au niveau de l'image et au niveau de la dcision.     Pour mieux apprhender la complexit de la dcision et rester dans un schma gnral     de rsolution,  la fois des caratristiques individuelles (relatives  une particule)     et globales (sur l'ensemble de l'image) sont extraites de l'image. Pour fusionner ces     informations htrognes, deux approches hirarchiques de fusion ont t     dveloppes. La premire est base sur des concepts d'analyse de donnes. La     deuxime est base sur des concepts de fusion de capteurs o les dcisions issues de     chaque caractristique sont combines pour fournir la dcision. Les dcisions     qualitatives ne sont pas formalisables : les approches sont bases sur le concept de     l'apprentissage par l'exemple. A ce guide stratgique, s'ajoute une aide pour le choix     des outils les plus adapts  une application donne. Ce guide a t utilis avec     succs pour la conception de capteurs pour l'valuation des boulanges en minoterie et     pour l'valuation de la qualit des grains dans des machines agricoles. </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="152"><font size="1" face="Arial"><b>Abstract</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="152"><font size="2">A guide for the     conception of classification systems of granular product populations using image analysis     has been realized. A non-specialist user can build intelligent sensors able to appreciate     granular products as well as an expert. Works on images and decision are required to     associate images of granular products and a qualitative decision. For better apprehending     the complexity of the decision and being general, both individual and global     characteristic are extracted on the image. To fuse these heterogeneous informations, two     hierarchical approaches have been developed. The first is based on a data analysis concept     . The second is based on a sensor fusion system where the decisions given by each     characteristic are combined to give the final decision. The qualitative decisions are not     ruled based therefore the approaches are based on the training concept. A help for choice     of the most adapted tools is added to this strategic guide. This guide has been tested     successfully for the sensor conception of the evaluation of mills in a semolina and grain     quality in an agricultural machine. </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="77"><font size="1" face="Arial"><b>Mots cls</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="77"><font size="2">PRODUIT GRANULAIRE/     SEUILLAGE ADAPTIF/ BIOINDUSTRIE/AGGLOMERAT/TRAITEMENT D'IMAGES / RECONNAISSANCE DE FORME /     GRANULOMETRIE / ANALYSE GRANULOMETRIQUE / RECONNAISSANCE DE GRANULOMETRIE / CLASSIFICATION     / FUSION / CAPTEUR / AIDE A LA DECISION / BIOTECHNOLOGIE / INTELLIGENCE ARTIFICIELLE/     VISION/ CAPTEUR INTELLIGENT//MODELE RCE/HOPFIELD/METHODE CNNR/KOHONEN/RACQ/CLUSTER </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Diffusion</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">Diffusion tous publics     </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Cote</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">95/0180 </font></td>   </tr>   <tr>     <td width="15%" style="border-right: medium solid rgb(252,246,207)" height="21"></td>     <td height="21"></td>     <td height="21"></td>   </tr>   <tr>     <td width="15%" bgcolor="#FCF6CF" style="border-right: medium solid rgb(252,246,207)" height="35"><a name="BM9"></a><a href="pubvisivol.htm#BM8"><font size="2">article        prcdent</font></a></td>     <td bgcolor="#FCF6CF" height="35"><a href="pubvisivol.htm#BM10"><font size="2">article        suivant</font></a></td>     <td bgcolor="#FCF6CF" height="35"><a href="pubvisivol.htm#BM1"><font size="2">dbut        de la page</font></a></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="31"><font size="1" face="Arial"><b>Type de document</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="31"><font size="2">Article - Revue     scientifique  comit de lecture </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="32"><font size="1" face="Arial"><b>Titre</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="32"><font size="2">Reconnaissance des     agglomrats dans les images de produits granulaires en utilisant les statistiques et les     rseaux de neurones </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Title</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">Recognition of     overlapping particles in granular product images using statistics and neural networks </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Auteur</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">ROS F./GUILLAUME     S./RABATEL G./SEVILA F. </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="31"><font size="1" face="Arial"><b>Anne de publication</b></font><font size="2">     </font></td>     <td valign="top" colspan="2" width="90%" height="31"><font size="2">1995 </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Source</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">Food control, vol. 6,     n 1 </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Langue</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">ENG </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Page</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">p. 37-43 </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="92"><font size="1" face="Arial"><b>Rsum</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="92"><font size="2">L'analyse d'image peut     tre utilise pour caractriser des populations granulaires dans plusieurs procds     notamment dans les industries agro-alimentaires ou dans le machinisme agricole. De l'image     on peut extraire  la fois des paramtres individuels et globaux. Parce qu'ils biaisent     les mesures sur les paramtres individuels, les agglomrats doivent faire l'objet d'une     gestion spcifique et donc tre identifis. Cette connaissance peut tre ralise en     modlisant les techniques d'analyse de l'image (pour l'extraction des caractristiques),     statistiques (pour rduire l'information) et d'intelligence artificielle  base de     rseaux de neurones (pour prendre la dcision). </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="77"><font size="1" face="Arial"><b>Abstract</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="77"><font size="2">Image analysis can be     used to characterize granular populations in many processes in the food industry or in     agricultural engineering. Either global or individual parameters can be extracted from the     image. However, granular products may agglomerate on the image, bringing bias measurements     of individual parameters : products which agglomerate have to be recognized. This is done     by a combination of image analysis (to pre-process and extract features), statistical     methods (to reduce information) and neural network techniques (to take decisions). </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="32"><font size="1" face="Arial"><b>Mots cls</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="32"><font size="2">PRODUIT     GRANULAIRE/TRAITEMENT D'IMAGES / CLASSIFICATION / INTELLIGENCE ARTIFICIELLE / RESEAU DE     NEURONES/ </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Diffusion</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">Diffusion tous publics     </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Cote</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">95/0079 </font></td>   </tr>   <tr>     <td width="15%" bgcolor="#FCF6CF" style="border-right: medium solid rgb(252,246,207)" height="35"><a name="BM10"></a><a href="pubvisivol.htm#BM9"><font size="2">article        prcdent</font></a></td>     <td colspan="2" bgcolor="#FCF6CF" height="35"><a href="pubvisivol.htm#BM1"><font size="2">dbut        de la page</font></a></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="31"><font size="1" face="Arial"><b>Type de document</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="31"><font size="2">Communication  un     congrs </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Titre</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">Evaluation de la     qualit des rcoltes : une approche par la granularit </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Title</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">Evaluation of yield     quality: a granularity approach </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="32"><font size="1" face="Arial"><b>Congrs</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="32"><font size="2">L'application de     l'analyse d'images dans l'agro-industrie : de la production  la transformation des     produits agricoles, Montpellier, 28 septembre 1994 </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Auteur</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">GUILLAUME S./ROS     F./FATOU J.M. </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="31"><font size="1" face="Arial"><b>Anne de publication</b></font><font size="2">     </font></td>     <td valign="top" colspan="2" width="90%" height="31"><font size="2">1994 </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Editeur</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">Cemagref Editions,     Antony </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Collection</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">Coll. Actes de     colloque </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Langue</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">FRE </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Page</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">p. 51-56 </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="107"><font size="1" face="Arial"><b>Rsum</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="107"><font size="2">Un capteur pour     caractriser les produits granulaires a t dvelopp. Il se compose de trois     sous-ensembles : un systme mcanique qui prlve une partie reprsentative de     l'chantillon  valuer, une camra CCD qui assure la prise d'images, un module     d'analyse d'images et de traitement des donnes. Le principe consiste  comparer un     chantillon  une &quot;classe de qualit&quot; pr-dfinie. Le systme de dcision     est construit par apprentissage supervis. Il est paramtr par un jeu d'exemples dont     on connat la classe de qualit. Ce capteur peut tre embarqu sur une machine du type     moissonneuse-batteuse et aider  l'apprciation de la qualit des crales en cours     de rcolte ou bien sur une ensileuse pour caractriser la qualit du hachage fin du     mas par exemple. </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="92"><font size="1" face="Arial"><b>Abstract</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="92"><font size="2">A sensor to     characterize granular products has been developed. It is made of three sub-groups: a     mechanical system which collects a representative part of the sample to be evaluated, a     CCD camera which takes the photos, an image processing and data treatment unit. The basic     idea is to compare a sample to a pre-defined &quot; quality class &quot;. The decision     system is built by supervized learning. It is adjusted using a set of examples of known     quality class. This sensor can be placed in a harvester-thresher type of machine and can     help appreciate grain quality during harvesting or on a silo filler to characterize the     grinding quality of corn, for example. </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="32"><font size="1" face="Arial"><b>Mots cls</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="32"><font size="2">PRODUIT     GRANULAIRE/CLASSE DE QUALITE/TRAITEMENT D'IMAGES / VISION ARTIFICIELLE / CAPTEUR /     CONTROLE DE QUALITE / CONTROLE AUTOMATIQUE / RECOLTE / CEREALE / CAMERA/ </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Diffusion</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">Diffusion tous publics     </font></td>   </tr>   <tr>     <td valign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="17"><font size="1" face="Arial"><b>Cote</b></font><font size="2"> </font></td>     <td valign="top" colspan="2" width="90%" height="17"><font size="2">94/0144 </font></td>   </tr>   <tr>     <td width="15%" bgcolor="#FCF6CF" style="border-right: medium solid rgb(252,246,207)" height="35"><a name="BM11"></a><font size="2">        <a href="pubvisivol.htm#BM10">article prcdent</a></font></td>     <td colspan="2" bgcolor="#FCF6CF" height="35"><a href="pubvisivol.htm#BM1"><font size="2">dbut        de la page</font></a></td>   </tr>   <tr>     <td vAlign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="35"><font face="arial" size="1"><b>Type de document</b></font> </td>     <td vAlign="top" colspan="2" width="90%" height="35">Communication  un congrs </td>   </tr>   <tr>     <td vAlign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="21"><font face="arial" size="1"><b>Titre</b></font> </td>     <td vAlign="top" colspan="2" width="90%" height="21">Application du traitement d'images     video  la classification de produits granulaires </td>   </tr>   <tr>     <td vAlign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="21"><font face="arial" size="1"><b>Title</b></font> </td>     <td vAlign="top" colspan="2" width="90%" height="21">Application of video image analysis     to the classification of granular products </td>   </tr>   <tr>     <td vAlign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="21"><font face="arial" size="1"><b>Congrs</b></font> </td>     <td vAlign="top" colspan="2" width="90%" height="21">SPIE Optics in agriculture forestery     and biological processing, Boston, USA, 2-4 November 1994 </td>   </tr>   <tr>     <td vAlign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="21"><font face="arial" size="1"><b>Auteur</b></font> </td>     <td vAlign="top" colspan="2" width="90%" height="21">ROS F./GUILLAUME S./BERTRAND     D./RABATEL G./SEVILA F. </td>   </tr>   <tr>     <td vAlign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="35"><font face="arial" size="1"><b>Anne de publication</b></font> </td>     <td vAlign="top" colspan="2" width="90%" height="35">1994 </td>   </tr>   <tr>     <td vAlign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="21"><font face="arial" size="1"><b>Langue</b></font> </td>     <td vAlign="top" colspan="2" width="90%" height="21">ENG </td>   </tr>   <tr>     <td vAlign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="21"><font face="arial" size="1"><b>Page</b></font> </td>     <td vAlign="top" colspan="2" width="90%" height="21">8 p. </td>   </tr>   <tr>     <td vAlign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="268"><font face="arial" size="1"><b>Rsum</b></font> </td>     <td vAlign="top" colspan="2" width="90%" height="268">Une approche mthodologique par     reconnaissance de forme est propose pour slectionner en ligne des produits     craliers quand ils sont dverss  l'air libre. Elle traite de la caractrisation     des populations de produits suivant diffrentes classes de qualit prdfinies par     analyse des images. Afin de tenir compte de la complexit de ce problme, nous vous     proposons d'utiliser les variables globales (relatives  l'image complte) et     individuelles (relatives  chaque particule). Le module de traitement des donnes est     destin  rduire la dimension de l'espace variable sans perdre d'informations puis de     slectionner les composants les plus pertinents pour former le systme de dcision. La     premire phase de ce systme, appele phase gnraliste, doit fournir une rponse     claire qui signifie slectionner un sous-ensemble de classes possibles. La seconde phase,     spcialise, qui est forme pour distinguer uniquement certains sous-ensembles de     classes fournie par la phase gnraliste, donne la dcision. La mthode a t     applique dans le cadre de la classification des produits de meunerie. Trois classes de     qualit ont t dfinies, elles correspondent aux cartements des premiers cylindres     de broyage (0,30, 0,40 et 0,50 mm) d'un broyeur pilote de semoule. Dans ces conditions, le     pourcentage de prcision de la classification obtenu  l'aide de ce systme est     suprieur  80 %. </td>   </tr>   <tr>     <td vAlign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="249"><font face="arial" size="1"><b>Abstract</b></font> </td>     <td vAlign="top" colspan="2" width="90%" height="249">A methodogical approach of pattern     recognition is proposed to make on-line selection of cereal products as they are poured     down through the air. It addresses the characterisation of product populations according     to different pre-defined quality classes using image analysis. In order to take into     account the complexity of this problem, we propose to use both global (related to the     whole image) and individual (related to each particle) variables. The aim of the data     processing module is to reduce the dimension of the variable space without losing     information, and then to select the most pertinent components to train the decision     system. The first stage of this system, called generalist one, has to give an ambiguous     response, that means to select a subset of possible output classes. The second, of     specialist, which is trained to distinguish only some subsets of classes delivered by the     generalist one, gives the decision. The method has been applied in the framework of     milling products classification. Three quality classes have been defined, they correspond     to the rolls gap (0.30, 0.40, 0.50 mm.) of the first break rolls of a semolina pilot mill.     In these conditions, the classification accuracy rate achieved by the system is higher     than 80 %. </td>   </tr>   <tr>     <td vAlign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="40"><font face="arial" size="1"><b>Mots cls</b></font> </td>     <td vAlign="top" colspan="2" width="90%" height="40">PRODUIT GRANULAIRE/ SELECTION EN     LIGNE/RECONNAISSANCE DE FORME / CONTROLE DE QUALITE / CEREALE/ </td>   </tr>   <tr>     <td vAlign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="21"><font face="arial" size="1"><b>Diffusion</b></font> </td>     <td vAlign="top" colspan="2" width="90%" height="21">Diffusion tous publics </td>   </tr>   <tr>     <td vAlign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" height="21"><font face="arial" size="1"><b>Cote</b></font> </td>     <td vAlign="top" colspan="2" width="90%" height="21">94/0741 </td>   </tr>   <tr>     <td vAlign="top" width="15%" style="border-right: medium solid rgb(252,246,207)" bgcolor="#FCF6CF" height="17"><a href="pubvisivol.htm#BM11"><font size="2">article        prcdent</font></a></td>     <td vAlign="top" colspan="2" width="90%" bgcolor="#FCF6CF" height="17"><a href="pubvisivol.htm#BM1"><font size="2">dbut        de la page</font></a></td>   </tr> </table>  <p align="center">&nbsp;</p>  <hr>  <p><font size="2" face="Arial">(c) Cemagref, Mise  jour le </font><font size="2"><!--webbot bot="Timestamp" s-type="REGENERATED" s-format="%d/%m/%Y" startspan -->17/02/2003<!--webbot bot="Timestamp" endspan i-checksum="12529" --></font></p> </body> </html> 
