Search

This paper explores the potentiality of using the completed local binary pattern (CLBP) for the classification of an urbanized oasis area situated in southeastern Tunisia, in very high spatial resolution GeoEye imagery. To further improve the spatial information description derived by CLBP, which is successfully used in face recognition, we applied the multi-scale completed local binary pattern (MSCLBP) to classify the single-dated image. A supervised classification framework preceded by mean-shift segmentation is applied using the texture features alone and combined with the normalized difference vegetation index (NDVI). As the segmentation is a crucial step to obtain a good mapping, considerations are given to select the optimal combination of mean-shift input parameters, such as spatial radius and range radius. The results of this study indicate that MSCLBP outperforms the single-scale CLBP and the gray-level co-occurrence matrix (GLCM) descriptors in the task of expressing the classes of interests.

Image segmentation consists of partitioning the image into different objects of interest. For a biological image, the segmentation step is important to understand the biological process. However, it is a challenging task due to the presence of different dimensions for cells, intensity inhomogeneity, and clustered cells. The marker-controlled watershed (MCW) is proposed for segmentation, outperforming the classical watershed. Besides, the choice of markers for this algorithm is important and impacts the results. For this work, two foreground markers are proposed: kernels, constructed with the software Fiji and Obj.MPP markers, constructed with the framework Obj.MPP. The new proposed algorithms are compared to the basic MCW. Furthermore, we prove that Obj.MPP markers are better than kernels. Indeed, the Obj.MPP framework takes into account cell properties such as shape, radiometry, and local contrast. Segmentation results, using new markers and illustrated on real Drosophila dataset, confirm the good performance quality in terms of quantitative and qualitative evaluation.

The Decimal Descriptor Patterns (DDP) operator is a recent method of analyzing textures. Based on structural analysis of pattern distribution, the DDP presents an excellent description of the spatial structure of image texture. In this chapter, two successful applications of this operator are highlighted. The first one consists in Multiple Sclerosis (MS) Lesions detection from 3D Magnetic Resonance (MR) brain images tissue. The second application is 3D face analysis. Different experimentations conducted on a two well-know three-dimensional databases. For comparison, two best known operators of texture measures that are considered as references in image analysis are chosen: the Grey Level Co-occurrence Matrix (GLCM) and the Local Binary Patterns (LBP). Several tests of texture classification are performed in the same conditions using the classifier multiclass Support Vector Machines (SVM). The performance of the three operators is tested in front of illumination, inhomogeneity, contrast, and noise level variations with a various and a large number of textures. Experimental results show clearly the robustness of the DDP operator.

Segmentation of biological images is a challenging task, due to non convex shapes, intensity inhomogeneity and clustered cells. To address these issues, a new algorithm is proposed based on the B-spline level set method. The implicit function of the level set is modelled as a continuous parametric function represented with the B-spline basis. It is different from the discrete formulation associated with conventional level set. In this paper the proposed framework takes into account properties of biological images. The algorithm is applied to Drosophila images, and compared to conventional level set and Marker Controlled Watershed (MCW). Results show good performance in term of the DICE coefficient, for noisy and noiseless images.

To face ecological constraints and growing energy demand we should resort to renewable energy sources. Using solar energy as input source for Stirling engine is an interesting alternative. The objective of this paper is the study and the simulation of a small-scale solar Stirling engine generator. The simulation deals with modeling mechanical as well as electrical parts of the system. All the thermal, thermodynamic and mechanical aspects are considered in the engine part and the solar collector. The electrical part composed of a generator and a convertor. The efficiency of the studied system is of 38%.

In this paper, a novel algorithm for automatic fabric defect classification was proposed, based on the combination of a texture analysis method and a support vector machine SVM. Three texture methods were used and compared, GLCM, LBP, and LPQ. They were combined with SVM’s classifier. The system has been tested using TILDA database. A comparative study of the performance and the running time of the three methods was carried out. The obtained results are interesting and show that LBP is the best method for recognition and classification and it proves that the SVM is a suitable classifier for such problems. We demonstrate that some defects are easier to classify than others.

Land cover classification is an important topic in remote sensing. It provides useful information for environmental projects and ecosystem management. To succeed the mapping task, interest should be given to select a suitable classification algorithm. The recent advances in machine learning have shown the very great performances of the deep neural networks for many applications including the land cover mapping. The purpose of this work is to exploit the performance of Convolutional Neural Networks CNN to classify a GeoEye satellite image. Two CNN models with three and four convolution layers are tested to show the impact of the number of layers on the accuracy result. The two architectures outperform, with an overall accuracy more than 81%, the pixel based classification using Random Forest classifier.

Since 2015, the Sentinel 2 satellite provides a medium to high spatial resolution (10m-30m) images. For studying the land cover of Gabes area, located in the South- East of Tunisia, we exploited the 10 m bands of this satellite. We have tested the supervised classification with the SVM classifier. The classification is preceded by a segmentation step. The spectral data, the vegetation index and the texture metrics (GLCM) are used for training. The best Overall Accuracy OA (92, 12%) is obtained when all the used features are combined.

In this paper, we aim to study the problem of 3D face recognition. This problem becomes complex when we consider the human face expressions and different face occlusions. In this context is situated our work, we use the 3D UMB database to evaluate two powerful methods: the 3D Local Binary Pattern (3D-LBP) and the 3D Grey Level Co-occurrence Matrix (3D-GLCM). We tested also the combination of these two approaches. A training phase is necessary, to pass to the test phase which is followed by classification. We use a multiclass Support Vector Machines (SVM) for classification. The experimental results prove that in this classification problem 3D-LBP is the more suitable.

Lead sulphide (PbS) thin films represent a direct band gap IV–VI p-type semiconductor material (Eg= 0.41 eV) and have a good potential application in solar cells, infrared detector devices, sensors ...

This paper proposes the recognition and classification of three mean woven fabrics, twill, satin and plain. The proposed classifier is based on the texture analysis of woven fabric images for the recognition. In the pattern recognition phase, three methods are tested and compared: Gabor wavelet, Local Binary Pattern operators (LBP) and gray-level co-occurrence matrices (GLCM). Taking advantage of the difference between the woven fabric textures, we adopt a technique which is based on the texture of the images in the pattern recognition phase. For the classification phase we used a support vector machine (SVM) which we have proven is a suitable classifier for this type of problem The experimental results show that some of the studied methods are more compatible with this classification problem than others. Although it is the oldest method, GLCM always remains accurate (97.2 %).The fusion of the Gabor wavelet and GLCM give the best result (98%), but the GLCM have the better running time.

In this paper, the performance of two new promising operators for the analysis of 3D textures based on feature extraction is validated. The first operator is the decimal descriptor patterns from three orthogonal planes (DDP-TOP), a new feature descriptor that considers the co-occurrences on three orthogonal planes. The second operator is the grey level co-occurrence matrix from three orthogonal planes (GLCM-TOP) which is an extension of the 2D grey level co-occurrence matrix method. In order to classify the MR images of brain into healthy and diseased, several tests are performed in the same conditions of work using the classifier multiclass support vector machines (SVMs). The local binary pattern (LBP), a best known method of texture analysis is used for comparison. Using the DDP-TOP operator, excellent experimental results are obtained that prove the robustness of our approach with respect to the noise level and to different image contrasts.

In this paper, we study the problem of rotation invariant texture classifications. There are several methods in texture recognition problem, we compare three best known methods such us: Gabor wavelet filter, Local Binary Pattern operators (LBP) and co-occurrence matrix (GLCM). A multi-class Support Vector Machines (SVM) is used as a classifier. The three methods are evaluated based on two different databases: Brodatz and Outex to bring out a comparative study about the discrimination capabilities of those different families of texture classification methods. The experimental results show that some of the studied methods are more compatible with this classification problem than the others. The SVM classifier approve the running time of the algorithm of classification.

This paper proposes the classification of three mean woven textile fabrics. The classifier is based on the texture analysis of woven fabric images for the recognition.