HOME /

Publications

2020
Mokhtari Bilal, Melkemi Kamal E., Michelucci Dominique, Foufou Sebti: Optimizing Query Perturbations to Enhance Shape Retrieval. In: Mathematical Aspects of Computer and Information Sciences. Vol. LNCS 11989. Springer ; 2020. pp. 422-437. Publisher's VersionAbstract
3D Shape retrieval algorithms use shape descriptors to identify shapes in a database that are the most similar to a given key shape, called the query. Many shape descriptors are known but none is perfect. Therefore, the common approach in building 3D Shape retrieval tools is to combine several descriptors with some fusion rule. This article proposes an orthogonal approach. The query is improved with a Genetic Algorithm. The latter makes evolve a population of perturbed copies of the query, called clones. The best clone is the closest to its closest shapes in the database, for a given shape descriptor. Experimental results show that improving the query also improves the precision and completeness of shape retrieval output. This article shows evidence for several shape descriptors. Moreover, the method is simple and massively parallel.
2019
Aldjia Boucetta, Kamal Eddine Melkemi: Score Level Fusion of Palmprint, Face and Iris Using Adaptive PSO. . Intenational Journal of Applied Metaheuristic Computing [Internet]. 2019;10 (3) :175-194. Publisher's VersionAbstract
Systems that use unimodal biometrics often suffer from various drawbacks such as noise in sensed data, variations that are due to intra class, nonuniversality, spoof attacks, restricted degrees of freedom and high error rates. These limitations can be solved effectively by combining two or more biometric modalities. In this article, a multimodal biometric fusion system is presented that combines palmprint, face and iris traits. The biometric fusion is performed at the score level in order to improve the accuracy of the system. Scores obtained from the three classifiers are fused using adaptive particle swarm optimization (PSO). The PSO use new multi objective fitness function. This function has two objectives, improve the recognition rate and reduce the total equal error rates. The experimental results of the proposed method achieve a recognition accuracy of 100%, with EER of 0.00%, using Gabor filter combined with dimensionality reduction techniques PCA, LDA and KFA. Experimental results show that multimodal biometric systems are much more accurate than the unimodal counterparts.
2017
A Benali, G Luque, E Alba and Kamal E. Melkemi: An improved problem aware local search algorithm for the DNA fragment assembly problem. Soft Computing. , Springer [Internet]. 2017;21 (7) :1709-1720. Publisher's VersionAbstract
DNA fragment assembly is a critical and essential early task in a genome project. This task leads to an NP-hard combinatorial optimization problem, and thus, efficient approximate algorithms are required to tackle large problem instances. The Problem Aware Local Search (PALS) is one of the most efficient heuristics for this problem in the literature. PALS gives fairly good solutions but the probability of premature convergence to local optima is significant. In this paper, we propose two modifications to the PALS heuristic in order to ameliorate its performance. The first modification enables the algorithm to improve the tentative solutions in a more appropriate and beneficial way. The second modification permits a significant reduction in the computational demands of the algorithm without significant accuracy loss. Computational experiments confirm that our proposals lead to a more efficient and robust assembler, improving both accuracy and efficiency.
Bilal Mokhtari, Kamal Eddine Melkemi, Dominique Michelucci, Sebti Foufou: A 3D shape matching and retrieval approach based on fusion of curvature and geometric diffusion features. International Journal of Computer Applications in Technology [Internet]. 2017;55 (2) :79-91. Publisher's VersionAbstract

The majority of shape matching and retrieval methods use only one single shape descriptor. Unfortunately, no shape descriptor is sufficient to provide suitable results for all kinds of shapes. The most common way to improve the performance of shape descriptors is to fuse them. In this paper, we propose a new 3D matching and retrieval approach based on a fully unsupervised fusion of curvature and geometric diffusion descriptors. In fact, to improve retrieval precision, we use two descriptors based on local and global features extracted from a shape, and automatically combine these features using a fusion method called Product rule. The Product rule combines values assigned to vertices by the two descriptors. This fusion rule gives better results compared to other well-known fusion schemes such as Max, Min and Linear rules. The proposed approach improves considerably the retrieval precision even with pose changes. This is shown through the retrieval results obtained on several popular 3D shape benchmarks.

Keywords: 3D shape matching, 3D shape retrieval, curvature-based features, diffusion geometry, combination schemes, feature fusion, 3D shape descriptors, triangular meshes, similarity measures, features extraction
Bilal Mokhtari, Kamal Eddine Melkemi, Dominique Michelucci, Sebti Foufou: Unsupervised geodesic convex combination of shape dissimilarity measures,. Pattern Recognition Letters, Elsevier [Internet]. 2017;98 (15) :46-52. Publisher's VersionAbstract
Dissimilarity or distance metrics are the cornerstone of shape matching and retrieval algorithms. As there is no unique dissimilarity measure that gives good performances in all possible configurations, these metrics are usually combined to provide reliable results. In this paper we propose to compute the best linear convex, or weighted, combination of any set of measured shape distances to enhance shape matching algorithms. To do so, a database is represented as a graph, where nodes are shapes and the edges carry the convex combination of dissimilarity measures. Weights are chosen to maximize the weighted distances between the query shape and shapes in the database. The optimal weights are solutions of a linear programming problem. This fully unsupervised method improves the outcomes of any set of shape similarity measures as shown in our experimental results performed on several popular 3D shape benchmarks.
2016
Ahmed Ben Said, Rachid Hadjidj, Kamal Eddine Melkemi, Sebti Foufou: Multispectral image denoising with optimized vector non-local mean filter. Digital Signal Processing, Elsevier [Internet]. 2016;58 :115-126. Publisher's VersionAbstract
Nowadays, many applications rely on images of high quality to ensure good performance in conducting their tasks. However, noise goes against this objective as it is an unavoidable issue in most applications. Therefore, it is essential to develop techniques to attenuate the impact of noise, while maintaining the integrity of relevant information in images. We propose in this work to extend the application of the Non-Local Means filter (NLM) to the vector case and apply it for denoising multispectral images. The objective is to benefit from the additional information brought by multispectral imaging systems. The NLM filter exploits the redundancy of information in an image to remove noise. A restored pixel is a weighted average of all pixels in the image. In our contribution, we propose an optimization framework where we dynamically fine tune the NLM filter parameters and attenuate its computational complexity by considering only pixels which are most similar to each other in computing a restored pixel. Filter parameters are optimized using Stein's Unbiased Risk Estimator (SURE) rather than using ad hoc means. Experiments have been conducted on multispectral images corrupted with additive white Gaussian noise. PSNR and similarity comparison with other approaches are provided to illustrate the efficiency of our approach in terms of both denoising performance and computation complexity.
2015
B. Mokhtari, Dominique Michelucci, Kamal E. Melkemi, Sebti Foufou, Proposition d’un descripteur de formes pour la recherche et la classification des maillages 3D, in Journées du Groupe de Travail en Modélisation Géométrique 2015, du Mercredi 1er au Jeudi 2 avril 2015, Université de Poitiers, France. Poitiers, France ; 2015. Publisher's VersionAbstract

Dans cet article, nous proposons une méthode pour reconnaître et apparier les formes 3D.

Les formes sont décrites par des maillages triangulaires. Nous utilisons le caractère localement saillant des sommets du maillage pour définir un premier descripteur S1, qui est attaché à chaque sommet. Nous utilisons aussi des propriétés plus globales de la «géométrie de la diffusion» – par exemple la diffusion de la chaleur – pour définir un second descripteur, S2, attaché lui aussi à chacun des sommets. Nous fusionnons ensuite ces deux descripteurs S1 et S2 en un seul, S, par la règle dite du produit. C’est théoriquement la fusion la plus simple et la plus efficace dans notre cas; nous le confirmons expérimentalement en comparant les résultats avec ceux obtenus avec les autres règles classiques de fusion de descripteurs : la somme, le minimum, le maximum, la moyenne. Nous démontrons la puissance de notre méthode en la comparant avec d’autres déjà proposées dans la littérature. Nous montrons que le descripteur S est invariant par transformation rigide, par changement de posture (assis, debout, couché, accroupi, avec ou sans les mains sur les hanches), et même par changement de topologie, etc. Par exemple, le descripteur S distingue les femmes (Victoria) des hommes (David).

2014
Bilal Mokhtari, Kamal Eddine Melkemi, Dominique Michelucci and Sebti Foufou DYNAMIC CLUSTERING-BASED METHOD FOR SHAPE RECOGNITION AND RETRIEVAL. In: Proceedings of TMCE 2014, May 19–23, 2014, Budapest, Hungary, edited by I. Horvath and Z. Rusakc , Organizing Committee of TMCE 2014, ISBN 978-94-6186-177-1. Budapest, Hungary: TMCE ; 2014. Publisher's VersionAbstract

This paper presents a shape matching framework based on a new shape decomposition approach. A new region-based shape descriptor is proposed to compute the best match between given 2D or 3D shapes. In order to find similar shapes in a database, we first split the interior of each shape into the adequate set of parts, classes, or ellipsoids, then find the corresponding parts between different shapes, and finally compute their similarity. Essentially, we compute the best shape decomposition into k classes using an improved version of the k-means clustering algorithm without prior fixing of the number of parts. Additionally, we propose a new tool which determines the best ellipsoids packing in order to efficiently represent a shape according to its components. The shape recognition process compares the optimal ellipsoidal partition of the new shape with the different models of a database and extracts the closest shapes. The performances of our shape matching framework are shown through experiments on various data of MPEG-7 and benchmark databases.

Khalil M. Mezghiche, Kamal E. Melkemi, Sebti Foufou: Matching with quantum genetic algorithm and shape contexts. In: 2014 IEEE / ACS 11th International Conference on Computer Systems and Applications (AICCSA). Doha, Qatar: IEEE ; 2014. pp. 536-542. Publisher's VersionAbstract
In this paper, we propose to combine the shape context (SC) descriptor with quantum genetic algorithms (QGA) to define a new shape matching and retrieval method. The SC matching method is based on finding the best correspondence between two point sets. The proposed method uses the QGA to find the best configuration of sample points in order to achieve the best possible matching between the two shapes. This combination of SC and QGA leads to a better retrieval results based on our tests. The SC is a very powerful discriminative descriptor which is translation and scale invariant, but weak against rotation and flipping. In our proposed quantum shape context algorithm (QSC), we use the QGA to estimate the best orientation of the target shape to ensure the best matching for rotated and flipped shapes. The experimental results showed that our proposed QSC matching method is much powerful than the classic SC method for the retrieval of shapes with orientation changes.
2013
B. Mokhtari, Dominique Michelucci, Kamal Eddine Melkemi, Sebti Foufou, Reconnaissances de formes par squelettes d’ellipsoïdes, in Journées du Groupe de Travail en Modélisation Géométrique 2013, du 27-28 mars 2013, Université de Marseille, France. . Marseilles ; 2013. Publisher's Version
Aldjia Boucetta, Kamal Eddine Melkemi: Hand shape recognition using Hu and Legendre moments. In: SIN '13: Proceedings of the 6th International Conference on Security of Information and Networks. New York NY United States: Association for Computing Machinery (ACM) ; 2013. Publisher's VersionAbstract
Hand recognition is one of an important biometric recognition technique. In this paper, we have proposed a simple and efficient technique for hand recognition that combines Hu invariant moments and Legendre moments.
2010
Kamal E. Melkemi, Sebti Foufou: Fuzzy Distributed Genetic Approaches for Image Segmentation. CIT Journal of Computing and Information Technology [Internet]. 2010;8 (3) : 221–231. Publisher's VersionAbstract
This paper presents a new image segmentation algorithm (called FDGA-Seg) based on a combination of fuzzy logic, multiagent systems and genetic algorithms. We propose to use a fuzzy representation of the image site labels by introducing some imprecision in the gray tones values. The distributivity of FDGA-Seg comes from the fact that it is designed around a MultiAgent System (MAS) working with two different architectures based on the master-slave and island models. A rich set of experimental segmentation results given by FDGA-Seg is discussed and compared to the ICM results in the last section.
2007
Sihem Slatnia, Mohamed Batouche, Kamal E. Melkemi, Evolutionary Cellular Automata Based-Approach for Edge Detection . In: Applications of Fuzzy Sets Theory, 7th International Workshop on Fuzzy Logic and Applications, WILF 2007, Camogli, Italy, July 7-10, 2007, Proceedings. Vol. LNCS. Camogli, Italy: Springer ; 2007. pp. 404-411.Abstract
We use an evolutionary process to seek a specialized powerful rule of Cellular Automata (CA) among a set of best rules for extracting edges in a given black-white image. This best set of local rules determines the future state of CA in an asynchronous way. The Genetic Algorithm (GA) is applied to search the best CA rules that can realize better the edge detection.
2006
Kamal E. Melkemi, Mohamed Batouche, Sebti Foufou: A multiagent system approach for image segmentation using genetic algorithms and extremal optimization heuristics. Pattern Recognition Letters, Elsevier [Internet]. 2006;27 (11) :1230-1238. Publisher's VersionAbstract
We propose a new distributed image segmentation algorithm structured as a multiagent system composed of a set of segmentation agents and a coordinator agent. Starting from its own initial image, each segmentation agent performs the iterated conditional modes method, known as ICM, in applications based on Markov random fields, to obtain a sub-optimal segmented image. The coordinator agent diversifies the initial images using the genetic crossover and mutation operators along with the extremal optimization local search. This combination increases the efficiency of our algorithm and ensures its convergence to an optimal segmentation as it is shown through some experimental results.
Kamal E. Melkemi, Mohamed Batouche, Sebti Foufou: A Multiresolution Approach Based on MRF and BakSneppen Models for Image Segmentation. Informatica, IOS press [Internet]. 2006;17 (2) :225-236. Publisher's VersionAbstract
The two major Markov Random Fields (MRF) based algorithms for image segmentation are the Simulated Annealing (SA) and Iterated Conditional Modes (ICM). In practice, compared to the SA, the ICM provides reasonable segmentation and shows robust behavior in most of the cases. However, the ICM strongly depends on the initialization phase. In this paper, we combine Bak–Sneppen model and Markov Random Fields to define a new image segmentation approach. We introduce a multiresolution technique in order to speed up the segmentation process and to improve the restoration process. Image pixels are viewed as lattice species of Bak–Sneppen model. The a-posteriori
2005
Kamal E. Melkemi, Mohamed Batouche, Sebti Foufou, MRF Model-Based Approach for Image Segmentation Using a Chaotic MultiAgent System. WILF 2005, LNCS Springer: . Milano, Italy. In: Fuzzy Logic and Applications, 6th International Workshop, WILF 2005, Crema, Italy, September 15-17, 2005, Revised Selected Papers. Vol. LNCS 3849. Springer. Milano, Italy: Springer ; 2005. pp. 344-353. Publisher's VersionAbstract
In this paper, we propose a new Chaotic MultiAgent System (CMAS) for image segmentation. This CMAS is a distributed system composed of a set of segmentation agents connected to a coordinator agent. Each segmentation agent performs Iterated Conditional Modes (ICM) starting from its own initial image created initially from the observed one by using a chaotic mapping. However, the coordinator agent receives and diversifies these images using a crossover and a chaotic mutation. A chaotic system is successfully used in order to benefit from the special chaotic characteristic features such as ergodic property, stochastic aspect and dependence on initialization. The efficiency of our approach is shown through experimental results.