Ezzeddine Zagrouba - Academia.edu (original) (raw)

Papers by Ezzeddine Zagrouba

Proceedings of the 16th International Joint Conference on e-Business and Telecommunications

Collusion presents a malicious attack for video watermarking techniques. In the case of anaglyph ... more Collusion presents a malicious attack for video watermarking techniques. In the case of anaglyph 3D video, this attack is not yet considered. In fact, only several watermarking techniques were proposed for this type of media and they are not robust against dangerous attacks such as MPEG compression and collusion. In this paper, a robust anaglyph 3D video watermarking technique is proposed. It is based on multi-sprites as a target of insertion. This allows obtaining a robustness against collusion attacks. First, several sprites are generated from original video. Then, a hybrid embedding scheme based on the least significant bit and the discrete wavelet transformation based method is applied on every sprite to insert signature. This improves invisibility and robustness against usual attacks. Experimental results show a high level of invisibility and a good robustness against collusion, compression and against additional attacks such as geometric and temporal attacks.

Proceedings of the 12th International Conference on Pattern Recognition Applications and Methods

2022 26th International Conference on Pattern Recognition (ICPR)

Keyframe extraction process consists on presenting an abstract of the entire video with the most ... more Keyframe extraction process consists on presenting an abstract of the entire video with the most representative frames. It is one of the basic procedures relating to video retrieval and summary. This paper present a novel method for keyframe extraction based on SURF local features. First, we select a group of candidate frames from a video shot using a leap extraction technique. Then, SURF is used to detect and describe local features on the candidate frames. After that, we analyzed those features to eliminate near duplicate keyframes, helping to keep a compact set, using FLANN method. We developed a comparative study to evaluate our method with three state of the art approaches based on local features. The results show that our method overcomes those approaches

Proceedings of the 13th International Conference on Agents and Artificial Intelligence

Nowadays, the integration of Cloud Computing and the Internet of Things (Cloud-IoT) has drawn att... more Nowadays, the integration of Cloud Computing and the Internet of Things (Cloud-IoT) has drawn attention as new technologies in the Future Internet. Cloud-IoT accommodates good solutions to address real-world problems by offering new services in real-life scenarios. Nonetheless, the traditional Cloud-IoT will be probably not going to give suitable service to the user as it handles enormous amounts of data at a single server. Furthermore, the Cloud-IoT shows huge security and privacy problems that must be solved. To address these issues, we propose an integrated Fog Cloud-IoT architecture based on Multi-Agents System and Blockchain technology. Multi-Agents System has proven itself in decision-making aspects, distributed execution, and its effectiveness in acting in the event of an intrusion without user intervention. On the other side, we propose Blockchain technology as a distributed, public, authentic ledger to record the transactions. The Blockchain represents a great advantage to the next generation computing to ensures data integrity and to allows low latency access to large amounts of data securely. We evaluated the performance of our proposed architecture and compared it with the existing models. The result of our evaluation shows that performance is improved by reducing the response time.

International Journal of Computer and Communication Engineering, 2017

In this paper, we propose a novel multimodal MRI and CT images fusion method based on Multi-resol... more In this paper, we propose a novel multimodal MRI and CT images fusion method based on Multi-resolution Singular Value Decomposition (MSVD) and Modified Pulse Coupled Neural Network (MPCNN).Firstly, the input pre-registered MRI and CT images are decomposed into high frequency (HF) and low frequency (LF) sub-bands by using the MSVD transform. Then, the MPCNN model is applied on each LF sub-bands. The proposed method can adaptively determine the linking strength of the MPCNN model. After that, LF coefficients are combined based on the output of MPCNN coefficients while HF coefficients are fused by using the maximum selection rule. Finally, the inverse MSVD is applied to reconstruct the fused image. Visual effect and objective evaluation criteria are used to evaluate the performance of our approach for nine pairs of MRI and CT images. The experimental results demonstrate that the proposed method has a better performance than other current methods.

IET Image Processing, 2020

The Visual Computer, 2021

Accurate glioma classification before surgery is of the utmost important in clinical decision mak... more Accurate glioma classification before surgery is of the utmost important in clinical decision making and prognosis prediction. In this paper, we investigate the impact of multi-modal MR image fusion for the differentiation of low-grade gliomas (LGG) versus high-grade gliomas (HGG) via integrative analyses of radiomic features and machine learning approaches. A set of 80 histologically confirmed gliomas patients (40 HGG and 40 LGG) obtained from the MICCAI BraTS 2019 data were involved in this study. To achieve this work, we propose to combine T1 with T2 or FLAIR modality in the non-subsampled shearlet domain. Firstly, the pre-processed source MR images are decomposed into low-frequency (LF) and several high-frequency (HF) sub-images. LF sub-images are fused using the proposed weight local features fusion rule while HF sub-images are combined based on the novel sum-modified-laplacian technique. Experimental results demonstrate that the proposed fusion approach outperformed the recent state-of-the-art approaches in terms of entropy and feature mutual information. Subsequently, a key radiomics signature was retrieved by the least absolute shrinkage and selection operator regression algorithm. Five machine learning classifiers were established and evaluated with the retrieved dataset, then with the fused dataset using tenfold cross-validation scheme. As a result, the random forest had the highest accuracy of 96.5% with 21 features selected from the raw data and 96.1% with 16 features selected from the fused data. Finally, the experimental findings confirm that the proposed aided diagnosis framework represents a promising tool to aid radiologists in differentiating HGG and LGG.

Eleventh International Conference on Machine Vision (ICMV 2018), 2019

In this paper, we investigate the classification of two soft tissue sarcoma subtypes within a mul... more In this paper, we investigate the classification of two soft tissue sarcoma subtypes within a multi-modal medical dataset based on three pre-trained deep convolutional networks of the ImageNet challenge. We use multiparametric MRI’s with histologically confirmed liposarcoma and leiomyosarcoma. Furthermore, the impact of depth on fine-tuning for medical imaging is highlighted. Therefore, we fine-tune the AlexNet along with deeper architectures of the VGG. Two configurations with 16 and 19 learned layers are fine-tuned. Experimental results reveal a 97.2% of classification accuracy with the AlexNet CNN, while better performance has been achieved using the VGG model with 97.86% and 98.27% on VGG-16-Net and VGG-19-Net, respectively. We demonstrated that depth is favorable for STS subtypes differentiation. Addionally, deeper CNN’s converge faster than shallow, despite, fine-tuned CNN‘s can be used as CAD to help radiologists in decision making.

Advances in Computational Intelligence, 2017

Trying to find clusters in high dimensional data is one of the most challenging issues in machine... more Trying to find clusters in high dimensional data is one of the most challenging issues in machine learning. Within this context, subspace clustering methods have showed interesting results especially when applied in computer vision tasks. The key idea of these methods is to uncover groups of data that are embedding in multiple underlying subspaces. In this spirit, numerous subspace clustering algorithms have been proposed. One of them is Sparse Subspace Clustering (SSC) which has presented notable clustering accuracy. In this paper, the problem of similarity measure used in the affinity matrix construction in the SSC method is discussed. Assessment on motion segmentation and face clustering highlights the increase of the clustering accuracy brought by the enhanced SSC compared to other state-of-the-art subspace clustering methods.

2017 IEEE/ACS 14th International Conference on Computer Systems and Applications (AICCSA), 2017

In this paper, a new region based active contour model is proposed for image segmentation. The pr... more In this paper, a new region based active contour model is proposed for image segmentation. The proposed model is based on the combination of an adaptive local term based on the computation of local statistics deduced at each point of the evolved curve and a global term built using the means of intensities inside and outside the evolved curve. The novelty of the approach is the introduction of an adaptive energy term by the definition of local regions along the curve that will be updated at each iteration of the minimization process according to the gradient information. Experiments on medical, synthetic, real-word and noisy images prove the effectiveness of the proposed method regarding methods of state of the art.

2017 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2017

Keyframe extraction is one of the basic procedures relating to video retrieval and summary. It co... more Keyframe extraction is one of the basic procedures relating to video retrieval and summary. It consists on presenting an abstract of the video with the most representative frames. This paper presents an efficient keyframe extraction approach based on local description and graph modularity clustering. The first step is to generate a set of candidate keyframes using a windowing rule in order to reduce the data to be examined. After that, detect interest points in these set of images. Then compute repeatability between each two images belonging to the candidate set and stocks these values in a matrix that we called repeatability matrix. Finally, the repeatability matrix is modelled by an oriented graph and we will select keyframes using graph modularity clustering principle. The experiments showed that this method succeeds in extracting keyframes while preserving the salient content of the video. Further, we found good values in term of precision, PSNR and compression rate.

3D Research, 2019

Thanks to the rapid growth of internet and the advanced development of 3D technology, 3D images a... more Thanks to the rapid growth of internet and the advanced development of 3D technology, 3D images and videos are proliferated over the networks. However, this causes several insecurity problems, and protecting this type of media has become a main challenge for many researchers. 3D watermarking is considered as an efficient solution for 3D data protection. In fact, it consists in embedding a secret key into a 3D content to protect it and in trying to extract it after any attack applied on marked 3D data. Anaglyph is the most popular and economical method among different 3D visualization methods. For this

Neural Computing and Applications, 2018

Recently, deep learning has been shown effectiveness in multimodal image fusion. In this paper, w... more Recently, deep learning has been shown effectiveness in multimodal image fusion. In this paper, we propose a fusion method for CT and MR medical images based on convolutional neural network (CNN) in the shearlet domain. We initialize the Siamese fully convolutional neural network with a pre-trained architecture learned from natural data; then, we train it with medical images in a transfer learning fashion. Training dataset is made of positive and negative patch pair of shearlet coefficients. Examples are fed in two-stream deep CNN to extract features maps; then, a similarity metric learning based on cross-correlation is performed aiming to learn mapping between features. The minimization of the logistic loss objective function is applied with stochastic gradient descent. Consequently, the fusion process flow starts by decomposing source CT and MR images by the non-subsampled shearlet transform into several subimages. High-frequency subbands are fused based on weighted normalized cross-correlation between feature maps given by the extraction part of the CNN, while lowfrequency coefficients are combined using local energy. Training and test datasets include pairs of pre-registered CT and MRI taken from the Harvard Medical School database. Visual analysis and objective assessment proved that the proposed deep architecture provides state-of-the-art performance in terms of subjective and objective assessment. The potential of the proposed CNN for multi-focus image fusion is exhibited in the experiments.

Multimedia Tools and Applications, 2019

Local Fisher Discriminant Analysis (LFDA) is a supervised feature extraction technique that prove... more Local Fisher Discriminant Analysis (LFDA) is a supervised feature extraction technique that proved to be efficient in reducing several types of data. However, it depends on the number of samples per class in a way that can lead, when classes are too large, to a consumption of all the memory of a commodity hardware, or to a disability to even run. To work around this limit, we hereby propose to introduce a parameter that adapts LFDA to the data's classes while accounting for the available resources on the used machine. In fact, according to this parameter, LFDA will consider a larger class as a set of smaller sub-classes and will process these latter instead of the larger one. We are calling our proposed optimization the classadapted LFDA, noted caLFDA. We also propose a Python implementation of LFDA and prove it more effective than the existent MATLAB implementation. To assess the efficiency of caLFDA, we applied it to reduce several hyperspectral images and compared the results of classifying the reduced images to the ones we get when using the original LFDA to reduce the data. When the hyperspectral images are too large for LFDA to be able to reduce them, we compare caLFDA's results to the ones we get with the most commonly used Principle Component Analysis (PCA).

IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2018

In this paper, a feature extraction approach based on a three-dimensional shearlet transform (she... more In this paper, a feature extraction approach based on a three-dimensional shearlet transform (shearlet 3D) is proposed. We aim at exploiting shearlet 3D to highlight the intrinsic properties of hyperspectral images (HSIs), well known by their correlated information and high dimensionality. First, we decompose the HSI to yield coefficients arranged in cubes that help the computing of statistical parameters. Afterward, using a simultaneous orthogonal matching pursuit (SOMP) algorithm, a classification process is carried out. SOMP relies on the powerful sparse representation paradigm, which helps representing data in a low-dimensional space. It is also built over the assumption that the contextual information incorporation into the sparse recovery problem improves the classification performance. To this algorithm, we propose to add an adapted decision rule where a similarity measurement is calculated to well assign the appropriate labels to pixels of interest. Experimental results proved that our proposed method outperformed state-of-the-art classifiers. Thanks to our proposed approach, we succeeded to build discriminative descriptors reaching high overall accuracies for two different HSI datasets, without taking into account all the shearlet 3D coefficients.

Journal of Electronic Imaging, 2016

The quest for optimal representations is considered a challenging goal in the field of image proc... more The quest for optimal representations is considered a challenging goal in the field of image processing. This consists of reducing the processing's complexity while ensuring an efficient reconstruction. An optimal representation should conserve the properties of the image pertaining to smooth content and contours. The multiscale geometric decompositions (MGD) were designed to reach this finality. They were used in many fields and for different purposes, such as feature extraction, detail enhancing, and change detection. A state-of-art of these decompositions is proposed in this paper. We present their theoretical definitions and how they capture the feature of the objects within an image. An overview table is elaborated where we summarize the methods, the data and the different criteria of assessment used in the studied cases. We are interested, particularly, in the use of MGD in a remote sensing (RS) context. Thus, some examples of their applications on RS images are studied. A discussion is presented based on the analyzed cases.

Lecture Notes in Computer Science, 2015

In this paper, we present an integer programming approach to estimating a discrete bi-colored ima... more In this paper, we present an integer programming approach to estimating a discrete bi-colored image from its two-color horizontal and vertical projections. The two-color projections basically refer to the number of pixels per column having colors c_1andandandc_2$$, and likewise for each row as well. The aim of the integer programming approach is to minimize the number of conflict pixels, i.e. the number of pixels that have color c_1aswellasas well asaswellasc_2$$. Since the problem is NP-complete, we give a survey of the literature and we propose a new integer programming formulation of this problem.

3D motion analysis by projecting trajectories on manifolds in a given video can be useful in diff... more 3D motion analysis by projecting trajectories on manifolds in a given video can be useful in different applications. In this work, we use two manifolds, Grassmann and Special Orthogonal group SO(3), to analyse accurately complex motions by projecting only skeleton data while dealing with rotation invariance. First, we project the skeleton sequence on the Grassmann manifold to model the human motion as a trajectory. Then, we introduce the second manifold SO(3) in order to consider the rotation that was ignored by the Grassmann manifold on the matched couples on this manifold. Our objective is to find the best weighted linear combination between distances in Grassmann and SO(3) manifolds according to the nature of the input motion. To validate the proposed 3D motion analysis method, we applied it in the framework of action recognition, re-identification and sport performance evaluation. Experiments on three public datasets for 3D human action recognition (G3D-Gaming, UTD-MHAD multimod...

In this paper, we are interested in comparing human trajectories using skeleton information provi... more In this paper, we are interested in comparing human trajectories using skeleton information provided by a consumer RGB-D sensor. In fact, 3D human joints given by skeletons offer an important information for human motion analysis. In this context, the use of manifolds has grown considerably in the computer vision community in recent years. The main contribution of this study resides in working jointly with two manifolds. The matching of the trajectories is performed in Stiefel manifold and dissimilarity measure is carried out in Grassmann manifold. Indeed, trajectories of motions are provided by the projection in the Stiefel manifold. Then, the Stiefel distance is used within the dynamic time warping in order to define the appropriate matching between a reference trajectory and a test one. This allows avoiding that the rotation within the motion will be ignored, as it is the case with the Grassmann manifold. Then, the dissimilarity is evaluated using the Grassmann distance to compar...