Abdelhani Boukrouche

One dimensional local binary pattern for bone texture characterization

The evaluation of osteoporotic disease from X-ray images presents a major challenge for pattern recognition and medical applications. Textured images from the bone microarchitecture of osteoporotic and healthy subjects show a high degree of similarity, thus drastically increasing the difficulty of classifying such textures. In this paper, we propose a new method to separate osteoporotic cases from healthy controls, using texture analysis. The idea consists in combining global and local information to better capture the image characteristics. Global information is characterized by image projection which conveys information about the global aspect of the texture. Local information is encoded by the local patterns using neighborhood operators. The proposed technique is based on the local binary pattern (LBP) descriptor which has been classically applied on two dimensional (2D) images. Our algorithm is a derived solution for the 1D projected fields of the 2D images. Experiments were conducted on two populations of osteoporotic patients and control subjects. Compared to the classical LBP, the proposed approach yields a better classification rate of the two populations.

Ear description and recognition using ELBP and wavelets

The human ear is a new technology in biometrics which is not yet used in a real context or in commercial applications. For this purpose of biometric system, we present an improvement for ear recognition methods that use Elliptical Local Binary Pattern operator as a robust technique for characterizing the fine details of the two dimensional ear images. The improvements are focused on feature extraction and dimensionality reduction steps. The realized system is mainly appropriate for verification mode; it starts by decomposing the normalized ear image into several blocks with different resolutions. Next, the textural descriptor is applied on each decomposed block. A problem of information redundancies is appeared due to the important size of the concatenated histograms of all blocks, which has been resolved by reducing of the histograms dimensionalities and by selecting of the pertinent information using Haar Wavelets. Finally, the system is evaluated on the IIT Delhi Database containing two dimensional ear images and we have obtained a success rate about 94% for 500 images from 100 persons.

System for automatic faces detection

The effectiveness of biometric authentication based on face mainly depends on the method used to locate the face in the image or video. This paper presents a hybrid system for faces detection, in a color image or video, in unconstrained cases, i.e. situations in which illumination, pose, occlusion and size of the face are uncontrolled. To do this, the new method of detection proposed in this system is based primarily on a technique of automatic learning by using the decision of three neural networks, a new method of feature extraction based on the principal of energy compaction in the DC coefficient using the discrete cosine transform and a technique of segmentation by skin color to reduce the space of research and to accelerate the process of detection. A whole of pictures (faces and no faces) are transformed to vectors of data which will be used for entrain the neural networks to separate between the two classes while the discrete cosine transform is used to reduce the dimension of the vectors, to eliminate the redundancies of information, and to store only the useful information in a minimum number of coefficients. The experimental results have showed that this hybridization of methods will gave a very significant improvement of the rate of the recognition, quality of detection and the time of execution.

Trabecular Bone Anisotropy Characterization Using 1D Local Binary Patterns

This paper presents a new method to characterize the texture of gray level bone radiographic images. The technique is inspired from the Local Binary Pattern descriptor which has been classically applied on two dimensional (2D) images. Our algorithm is a derived solution for the 1D projected fields of the 2D images. The method requires a series of preprocessing of images. A clinical study is led on two populations of osteoporotic and control patients. The results show the ability of our technique to better discriminate the two populations than the classical LBP method. Moreover, they show that the structural organization of bone is more anisotropic for the osteoporotic cases than that of the control cases in accordance with the natural evolution of bone tissue linked to osteoporosis.

Identity recognition based on the external shape of the human ear

External shape of the human ear presents a rich and stable information embedded on the curved 3D surface, which has invited lot attention from the forensic and engineer scientists in order to differentiate and recognize people. However, recognizing identity from external shape of the human ear in unconstrained environments, with insufficient and incomplete training data, dealing with strong person-specificity, and high within-range variance, can be very challenging. In this work, we implement a simple yet effective approach which uses and exploits recent local texture-based descriptors to achieve faster and more accurate results. Support Vector Machines (SVM) are used as a classifier. We experiment with two publicly available databases, which are IIT Delhi-1 and IIT Delhi-2, consisting of several ear benchmarks of different natures under varying conditions and imaging qualities. The experiments show excellent results beyond the state-of-the-art.

1DLBP and PCA for face recognition

A new algorithm for face recognition is proposed in this work, this algorithm is mainly based on LBP texture analysis in one dimensional space 1DLBP and Principal Component Analysis PCA as a technique for dimensionalities reduction. The extraction of the faces features is inspired from the principal that the human visual system combines between local and global features to differentiate between people. Starting from this assumption, the facial image is decomposed into several blocks with different resolution, and each decomposed block is projected in one dimensional space. Next, the proposed descriptor 1DLBP is applied for each projected block. Then, the resulting vectors will be concatenated in one global vector. Finley, Principal Component Analysis is used to reduce the dimensionalities of the global vectors and to keep only the pertinent information for each person. The experimental results applied on AR database have showed that the proposed descriptor 1DLBP combined with PCA have given a very significant improvement at the recognition rate and the false alarm rate compared with other methods of face recognition, and a good effectiveness against to deferent external factors as: illumination, rotations and noise.

Face recognition using 1DLBP, DWT and SVM

The popular Local binary patterns (LBP) have been highly successful in describing and recognizing faces. However, the original LBP has several limitations which must to be optimized in order to improve its performances to make it suitable for the needs of different types of problems. In this paper, we investigate a new local texture descriptor for automated human identification using 2D facial imaging, this descriptor, denoted: One Dimensional Local Binary Pattern (1DLBP), produces binary code and inspired from classical LBP. The performances of the textural descriptor have been improved by the introduction of the wavelets in order to reduce the dimensionalities of the resulting vectors without losing information. The 1DLBP descriptor is assessed in comparison to the classical and the extended versions of the LBP descriptor. The experimental results applied on two publically datasets, which are the ORL and AR databases, show that this proposed approach of feature extraction, based on 1DLBP descriptor, given very significant improvements at the recognition rates, superiority in comparison to the state of the art, and a good effectiveness in the unconstrained cases.

Ear recognition using local color texture descriptors from one sample image per person

Morphological shape of the human ear presents a rich and stable information embedded on the curved 3D surface, which has invited lot attention from the forensic and engineer scientists in order to differentiate and recognize people. However, recognizing identity from morphological shape of the human ear using one sample image per person in training-set, with insufficient and incomplete training data, dealing with strong person-specificity can be very challenging. To address such problem, we propose a simple yet effective approach which uses and exploits local color texture descriptors in order to achieve faster and more accurate results. Support Vector Machine (SVM) is used as a classifier. We experiment with USTB-1 database consisting of several RGB ear benchmarks of different natures taken under varying conditions and imaging qualities. The experiments show excellent results beyond the state-of-the-art.

Experiments and improvements of ear recognition based on local texture descriptors

Abstract. The morphology of the human ear presents rich and stable information embedded on the curved 3-D surface and has as a result attracted considerable attention from forensic scientists and engineers as a biometric recognition modality. However, recognizing a person’s identity from the morphology of the human ear in unconstrained environments, with insufficient and incomplete training data, strong person-specificity, and high within-range variance, can be very challenging. Following our previous work on ear recognition based on local texture descriptors, we propose to use anatomical and embryological information about the human ear in order to find the autonomous components and the locations where large interindividual variations can be detected. Embryology is particularly relevant to our approach as it provides information on the possible changes that can be observed in the external structure of the ear. We experimented with three publicly available databases, namely: IIT Delhi-1, IIT Delhi-2, and USTB-1, consisting of several ear benchmarks acquired under varying conditions and imaging qualities. The experiments show excellent results, beyond the state of the art.

Person identification based on ear morphology

Morphological shape of the human ear presents a rich and stable information embedded on the curved 3D surface, which has invited lot attention from the forensic and engineer scientists in order to differentiate and recognize people. However, recognizing identity from morphological shape of the human ear in unconstrained environments, with insufficient and incomplete training data, dealing with strong person-specificity, and high within-range variance, can be very challenging. In this work, we implement a simple yet effective approach which uses and exploits recent local texture-based descriptors to achieve faster and more accurate results. Support Vector Machine (SVM) is used as a classifier. We experiment with two publicly available databases, which are IIT Delhi-1 and IIT Delhi-2, consisting of several ear benchmarks of different natures under varying conditions and imaging qualities. The experiments show excellent results beyond the state-of-the-art.