Pierre Loonis

Combination, cooperation and selection of classifiers: a state of the art

When several classifiers are brought to contribute to the same task of recognition, various strategies of decisions, implying these classifiers in different ways, are possible. A first strategy consists in deciding using different opinions: it corresponds to the combination of classifiers. A second strategy consists in using one or more opinions for better guiding other classifiers in their training stages, and/or to improve the decision-making of other classifiers in the classification stage: it corresponds to the cooperation of classifiers. The third and last strategy consists in giving more importance to one or more classifiers according to various criteria or situations: it corresponds to the selection of classifiers. The temporal aspect of Pattern Recognition (PR), i.e. the possible evolution of the classes to be recognized, can be treated by the strategy of selection.

The fuzzy c+2-means: solving the ambiguity rejection in clustering

Abstract In this paper we deal with the clustering problem whose goal consists of computing a partition of a family of patterns into disjoint classes. The method that we propose is formulated as a constrained minimization problem, whose solution depends on a fuzzy objective function in which reject options are introduced. Two types of rejection have been included: the ambiguity rejection which concerns patterns lying near the class boundaries and the distance rejection dealing with patterns that are far away from all the classes. To compute these rejections, we propose an extension of the fuzzy c-means (FcM) algorithm of Bezdek, Pattern Recognition with Fuzzy Objective Function Algorithms, Plenum Press, New York, 1981. This algorithm is called the fuzzy c+2-means (Fc+2M). These measures allow to manage uncertainty due both to imprecise and incomplete definition of the classes. The advantages of our method are (1) the degree of membership to the reject classes for a pattern xk are learned in the iterative clustering problem; (2) it is not necessary to compute other characteristics to determine the reject and ambiguity degrees; (3) the partial ambiguity rejections introduce a discounting process between the classical FcM membership functions in order to avoid the memberships to be spread across the classes; (4) the membership functions are more immune to noise and correspond more closely to the notion of compatibility. Preliminary computational experiences on the developed algorithm are encouraging and compared favorably with results from other methods as FcM, FPcM and F(c+1)M (fuzzy c+1-means: clustering with solely distance rejection) algorithms on the same data sets. The differences in the performance can be attributed to the fact that ambiguous patterns are less accounted in for the computing of the centers.

Segmentation and retrieval of ancient graphic documents

The restoration and preservation of ancient documents is becoming an interesting application in document image analysis. This paper introduces a novel approach aimed at segmenting the graphical part in historical heritage called lettrine and extracting its signatures in order to develop a Content-Based Image Retrieval (CBIR) system. The research principle is established on the concept of invariant texture analysis (Co-occurrence and Run-length matrices, Autocorrelation function and Wold decomposition) and signature extraction (Mininum Spanning Tree and Pairwise Geometric Attributes). The experimental results are presented by highlighting difficulties related to the nature of strokes and textures in lettrine. The signatures extracted from segmented areas of interest are informative enough to gain a reliable CBIR system.

A fuzzy petri net for pattern recognition: application to dynamic classes

Abstract. When involving evolutionary natural objects, the odeling of dynamic lasses is the main issue for a pattern recognition system. This problem an be avoided by making dynamic the syste of pattern recognition which an then enter into various states according to the evolution of the lasses. We propose a dynamic recognition system founded on two types of learning. The static aspect of the learning is ensured by lassifiers or systems of lassifiers, while the dynamic aspect is translated by the learning of the planning of the various states by a fuzzy Petri net. The method is sucessfully applied to a synthetic data set.

A Topological Measure for Image Object Recognition

All the effective object recognition systems are based on a powerful shape descriptor. We propose a new method for extracting the topological feature of an object. By connecting all the pixels constituting the object under the constraint to define the shortest path (minimum spanning tree) we capture the shape topology. The tree length is in the first approximation the key of our object recognition system. This measure (with some adjustments) make it possible to detect the object target in several geometrical configurations (translation / rotation) and it seems to have many desirable properties such as discrimination power and robustness to noise, that is the conclusion of the preliminary tests on characters and symbols.

A multiple classifier system using ambiguity rejection for clustering-classification cooperation

This article aims at showing that supervised and unsupervised learnings are not competitive, but complementary methods. We propose to use a fuzzy clustering method with ambiguity rejection to guide the supervised learning performed by bayesian classifiers. This method detects ambiguous or mixed areas of a learning set. The problem is seen from the multi-decision point of view (i.e. several classification modules). Each classification module is specialized on a particular region of the feature space. These regions are obtained by fuzzy clustering and constitute the original data set by union. A classifier is associated with each cluster. The training set for each classifier is then defined on the cluster and its associated ambiguous clusters. The overall system is parallel, since different classifiers work with their own training data clusters. The algorithm makes possible the adaptive classifier selection in the sense that the fuzzy clustering with ambiguity rejection gives adapted training data regions of the feature space. The decision making is the fusion of outputs from the most adapted classifiers.

Interpretation of Sound Tomography Image for the Recognition of Ganoderma Infection Level in Oil Palm

Basal stem rot (BSR) disease in oil palm caused by a group of decaying fungi called Ganoderma is considered as the most serious disease faced by the oil palm plantations in SouthEast Asia [1]. Significant yield losses can be observed when the number of palms infected by the fungus increases in the plantation as the infected palms will produce less quality fruit and eventually die thus requiring an early replanting. Flood et al. [2] and Idris and Ariffin [1] reported that the disease spread through contact of roots starting with the contact of an exiting inoculum in the soil and later spreading to the other palms through contact between the infected root with roots of other palms. The infection then moved upwards in the palm stem. The disease normally concentrated in the lower 1 meter of the trunk. Ganoderma produces enzymes that will degrade the oil palm tissue and affect the infected oil palm xylem thus causing serious problems to the distribution of water and other nutrients to the top of the palm tree. Because oil palm stems have no vascular cambium, they are essentially devoid of secondary growth. Therefore, palms cannot repair injuries to their stems. As the infection happens inside the palm stem, no physical symptoms can be detected at the early stage of infection. Basidiocarp is the most identifiable structure associated with the fungus. The conk originates from the fungus that grows in the infected trunk. However, most of the time, the conk does not appear at the early stage of the infection, making early detection of the disease very difficult. The foliar symptoms of BSR are reported to be the most apparent visual sign of the infection. However, these symptoms can only be considered as symptoms of Ganoderma infection if they are accompanied by the development of basidiocarp. The problem is that by the time these symptoms appear, usually over half of the lower internal stem tissues have been killed by the fungus.

A NEW MINIMUM SPANNING TREE-BASED METHOD FOR SHAPE DESCRIPTION AND MATCHING WORKING IN DISCRETE COSINE SPACE

In this article, a new minimum spanning tree-based method for shape description and matching is proposed. Its properties are checked through the problem of graphical symbols recognition. Recognition invariance in front shift and multi-oriented noisy objects was studied in the context of small and low resolution binary images. The approach seems to have many desirable properties, even if the construction of graphs induces an expensive algorithmic cost. In order to reduce time computing, an alternative solution based on image compression concepts is provided. The recognition is realized in a compact space, namely the Discrete Cosine space. The use of block discrete cosine transform is discussed and justified. The experimental results led on the GREC2003 database show that the proposed method is characterized by a good discrimination power, a real robustness to noise, with an acceptable time computing. The position with a reference approach like Zernike moments is also investigated to measure the relevance of the proposed technique.

Quality controls for gamma cameras and PET cameras: development of a free open-source ImageJ program

Acquisition data and treatments for quality controls of gamma cameras and Positron Emission Tomography (PET) cameras are commonly performed with dedicated program packages, which are running only on manufactured computers and differ from each other, depending on camera company and program versions. The aim of this work was to develop a free open-source program (written in JAVA language) to analyze data for quality control of gamma cameras and PET cameras. The program is based on the free application software ImageJ and can be easily loaded on any computer operating system (OS) and thus on any type of computer in every nuclear medicine department. Based on standard parameters of quality control, this program includes 1) for gamma camera: a rotation center control (extracted from the American Association of Physics in Medicine, AAPM, norms) and two uniformity controls (extracted from the Institute of Physics and Engineering in Medicine, IPEM, and National Electronic Manufacturers Association, NEMA, norms). 2) For PET systems, three quality controls recently defined by the French Medical Physicist Society (SFPM), i.e. spatial resolution and uniformity in a reconstructed slice and scatter fraction, are included. The determination of spatial resolution (thanks to the Point Spread Function, PSF, acquisition) allows to compute the Modulation Transfer Function (MTF) in both modalities of cameras. All the control functions are included in a tool box which is a free ImageJ plugin and could be soon downloaded from Internet. Besides, this program offers the possibility to save on HTML format the uniformity quality control results and a warning can be set to automatically inform users in case of abnormal results. The architecture of the program allows users to easily add any other specific quality control program. Finally, this toolkit is an easy and robust tool to perform quality control on gamma cameras and PET cameras based on standard computation parameters, is free, run on any type of computer and will soon be downloadable from the net (http://rsb.info.nih.gov/ij/plugins or http://nucleartoolkit.free.fr).

Fusion of heterogeneous and noisy informations: application to the quantification of the coronary stenosis

Proposes an algorithm to evaluate the similarity between two space-time distributions. One is obtained by experiment; the other is estimated by a numerical calculus. These are heterogeneous information types; their locations, as well as their densities and their reliabilities, are various. We have developed a first method which is correct when numerical and experimental information are closely linked. Nevertheless, in real-world problems, the initial conditions which induce the numerical information are vague. For the nonlinearity of the studied phenomena, the degree of similarity of both types of information is deeply degraded. Our approach is robust relative to this noise. It is used as part of a coronary stenosis identification process.