Anis Ladgham

A New Images Segmentation Method Based on Modified Particle Swarm Optimization Algorithm

The partitioning of an image into several constituent components is called image segmentation. Many approaches have been developed; one of them is the particle swarm optimization (PSO) algorithm, which is widely used. PSO algorithm is one of the most recent stochastic optimization strategies. In this article, a new efficient technique for the magnetic resonance imaging (MRI) brain images segmentation thematic based on PSO is proposed. The proposed algorithm presents an improved variant of PSO, which is particularly designed for optimal segmentation and it is called modified particle swarm optimization. The fitness function is used to evaluate all the particle swarm in order to arrange them in a descending order. The algorithm is evaluated by performance measures such as run time execution and the quality of the image after segmentation. The performance of the segmentation process is demonstrated by using a defined set of benchmark images and compared against conventional PSO, genetic algorithm, and PSO with Mahalanobis distance based segmentation methods. Then we applied our method on MRI brain image to determinate normal and pathological tissues.

Fast MR brain image segmentation based on modified Shuffled Frog Leaping Algorithm

Due to the need of correct diseases analysis, MR image segmentation remains till now a challenging problem, especially in the presence of random noise. This paper proposes a new meta-heuristic algorithm for MR brain image segmentation, named Modified Shuffled Frog Leaping Algorithm (MSFLA), based on the technique of Shuffled Frog Leaping Algorithm (SFLA). In this new paradigm, there is no need to filter the original image. The new fitness function proposed in our algorithm helps to evaluate quickly the particle frogs in order to arrange them in descending order. The proposed approach has been compared with other meta-heuristics such as 3D-Otsu thresholding with SFLA and Genetic Algorithm (GA) and also with the algorithm of segmentation using the Rician Classifier (RiCE). Experimental results show that the proposed MSFLA is able to achieve better segmentation quality and execution time than the latest methods.

Real Time Implementation of Detection of Bacteria in Microscopic Images Using System Generator

This paper outlines efficient hardware architecture of detection of bacteria and alga in microscopic images, using Xilinx System Generator (XSG). XSG is a high-level design tool based on blocks. It gives bit and cycle accurate simulation. The approach of detection used is the Hough transform. The latter is a very efficient approach of location of parametric curves in an image, especially lines. System was implemented on Virtex-V FPGA. To demonstrate the quality of the system, some experiments on microscopic images are given.

A 3D–2D face recognition method based on extended Gabor wavelet combining curvature and edge detection

The main limitation in 3D face recognition (FR) systems is their susceptibility to scanning difficulties and uncontrolled environments such as pose, illumination and expression variety. This paper proposes a new FR framework based on 3D to 2D mesh deforming and combined Gabor curvature and edge maps. The advantage of this method comes from the powerful saliency distribution achieved from applying extended Gabor wavelets to 2D projected face meshes. The extracted feature vectors are classified using the outstanding robustness of the support vector machine. Experiments carried out on common databases proved that valid accuracy rates can be accomplished by the proposed approach comparing to other existing methods.

MRI brain tumor recognition using Modified Shuffled Frog Leaping Algorithm

This paper presents a novel optimal algorithm for MRI brain tumor recognition. To do this, we use the newly developed meta-heuristic MSFLA (Modified Shuffled Frog Leaping Algorithm). Otherwise, a suitable choice of the fitness function ensures faster time of research with greater chance of convergence to the optimal value. The calculation of the used fitness function is linked to the image. The image must be scanned to calculate this function. For this, this function assists to quickly discover the adequate area modeling the tumor. Computer simulation results illustrate the effectiveness of the developed algorithm.

A novel optimised face recognition application based on modified shuffled frog leaping algorithm

In this work, we bring to light a novel face recognition (FR) system based on modified shuffled frog leaping algorithm (MSFLA) blended to Gabor wavelets. This new approach operates straightly on feature extraction and selection stages by providing the most propitious Gabor representations of a face image. While many researchers are seeking to find better parameterisation for Gabor filters, we introduce our evolutionary MSFLA-Gabor prototype combined to support vector machine (SVM) classifier as a robust contribution in the face biometric field. Primarily, we start by highlighting the impressive quality insured by Gabor filters in salient point extraction. Next, we present the potential dynamism of metaheuristic MSFLA in enhancing feature selection as well as up grading SVM classifier performance. Then, our optimised MSFLA-Gabor-SVM algorithm is tested on three databases under varied facial expressions, illuminations and poses. The experimental results have shown higher recognition rates and lower computational complexity scores than previous techniques.

Enhanced 3D mesh for face recognition

We propose a robust face identification method based on high saliency extraction. The adopted algorithm is performed on enhanced tri-dimensional data mesh. The enhancement stage aims to upgrade the quality of the scanned data by inhibiting noises, correcting missing information and smoothing the surface. Then, Gaussian curvatures and mean curvatures are calculated from principal curvature computation in furtherance of feature extraction. Next, referential curvature points are selected and utilized to perform the matching process. The results of our experimental essay have been evaluated by comparing them to similar advanced studies and have advantageously manifested bright levels of identification rates.

Gabor-SVM applied to 3D-2D deformed mesh model

We propose a robust method for 3D face recognition using 3D to 2D modeling and facial curvatures detection. The 3D-2D algorithm permits to transform 3D images into 3D triangular mesh, then the mesh model is deformed and fitted to the 2D space in order to obtain a 2D smoother mesh. Then, we apply Gabor wavelets to the deformed model in order to exploit surface curves in the detection of salient face features. The classification of the final Gabor facial model is performed using the support vector machines (SVM). To demonstrate the quality of our technique, we give some experiments using the 3D AJMAL faces database. The experimental results prove that the proposed method is able to give a good recognition quality and a high accuracy rate.

Multi-level fractional order PSO new paradigm algorithm for image segmentation

Nowadays, the resolution of the image characterisation problem is a very active and developed research field. Various applications areas were processed such as robotics and autonomous systems, computer vision, map processing and medical imaging. Pre-processing, segmentation, recognition and classification are the main areas of study. Multilevel segmentation of Benchmark images is our application field in this study. It is used to separate the original image into regions with common characteristics of an interesting viewing quality and a fast time execution. The purpose is to automatically determine the optimal threshold values based between-class variance maximisation. The choice of the already used method essentially depends on the quantitative characteristics. We note the optimal threshold values, the minimum CPU processing time and the stability of the proposed fitness function. Likewise, qualitatively results are required. The principle aim of this paper is to propose a new PSO algorithm for multilevel segmentation based on a novel fitness function and modified inertia component to find the optimal thresholds. Experimental results applied on a set of benchmarks images have been proven efficiencies and advantages in multilevel compared to other metaheuristics such as Genetic Algorithms (GA), Otsu method, Conventional PSO and Fractional-Order Darwinian PSO (DPSO).

Hardware implementation of Gabor filter for MRI using XSG

Feature extraction from images is an important stage for a recognition system. In this paper, we propose hardware architecture for the extraction of textural features from Magnetic Resonance Imaging (MRI) based on Gabor filter. The strength of Gabor filters in image processing is the similarity of their frequency and orientation representations to those of the human visual system. They are appropriate for texture representation and discrimination. The present architecture is developed using the Xilinx System Generator (XSG) and is implemented on Virtex-V Field Programmable Gate Array (FPGA). The implementation of our method on FPGA gives it the advantages of portability and real time extraction. The performances of the proposed method are demonstrated using a set of medical images.