A5022042532- Medical Image Segmentation Techniques
- Brain Tumor Detection and Classification
- Image Retrieval and Classification Techniques
- Image Enhancement Techniques
- Advanced Image and Video Retrieval Techniques
- Advanced Image Fusion Techniques
- Image and Signal Denoising Methods
- Domain Adaptation and Few-Shot Learning
- Advanced Neural Network Applications
- Topological and Geometric Data Analysis
- Dementia and Cognitive Impairment Research
- Digital Imaging for Blood Diseases
- Privacy, Security, and Data Protection
- AI in cancer detection
- Advanced Malware Detection Techniques
- Functional Brain Connectivity Studies
- Blockchain Technology Applications and Security
- Vehicular Ad Hoc Networks (VANETs)
- Advanced Neuroimaging Techniques and Applications
- Advanced Vision and Imaging
- Advanced Image Processing Techniques
- Network Security and Intrusion Detection
- Alzheimer's disease research and treatments
- Remote-Sensing Image Classification
- Image Processing Techniques and Applications
Edge Hill University
2021-2024
Laboratoire Bordelais de Recherche en Informatique
2012-2021
Institut Polytechnique de Bordeaux
2010-2021
Université de Bordeaux
2010-2021
Centre National de la Recherche Scientifique
2008-2021
University of Central Lancashire
2016-2021
Institut national de recherche en informatique et en automatique
2014-2015
Université Claude Bernard Lyon 1
2015
Institut National des Sciences Appliquées de Lyon
2014
Budapest University of Technology and Economics
2014
Abstract Whole brain segmentation of fine-grained structures using deep learning (DL) is a very challenging task since the number anatomical labels high compared to available training images. To address this problem, previous DL methods proposed use single convolution neural network (CNN) or few independent CNNs. In paper, we present novel ensemble method based on large CNNs processing different overlapping areas. Inspired by parliamentary decision-making systems, propose framework called...
In this paper, we address the problem of recovering a color image from grayscale one. The input data comes source considered as reference image. Reconstructing missing pixel is here viewed automatically selecting best among set candidates while simultaneously ensuring local spatial coherency reconstructed information. To solve problem, propose variational approach where specific energy designed to model selection and constraint problems simultaneously. contributions paper are twofold. First,...
In a world where organisations are embracing new IT working models such as Bring Your Own Device (BYOD) and remote working, the traditional mindset of defending network perimeter is no longer sufficient. Zero Trust Architecture (ZTA) has recently emerged security model in which breach dominates threat model. By default, ZTA considers any endpoint (i.e., device), user, or application to be untrusted until proven otherwise. Nonetheless, once by endpoint, using Advanced Persistent Threats...
Mathematical morphology (MM) offers a wide range of operators to address various image processing problems. These can be defined in terms algebraic (discrete) sets or as partial differential equations (PDEs). In this paper, we introduce nonlocal PDEs-based morphological framework on weighted graphs. We present and analyze set that leads family discretized PDEs Our formulation introduces patch-based configurations for extends approach the arbitrary data such nonuniform high dimensional data....
Abstract Numerous studies have proposed biomarkers based on magnetic resonance imaging (MRI) to detect and predict the risk of evolution toward Alzheimer’s disease (AD). Most these methods focused hippocampus, which is known be one earliest structures impacted by disease. To date, patch-based grading approaches provide among best hippocampus. However, this structure complex divided into different subfields, not equally AD. Former in - vivo mainly investigated structural alterations subfields...
Superpixels have become very popular in many computer vision applications. Nevertheless, they remain under-exploited, since the superpixel decomposition may produce irregular and nonstable segmentation results due to dependency image content. In this paper, we first introduce a novel structure, superpixel-based patch, called SuperPatch. The proposed based on neighborhood, leads robust descriptor, spatial information is naturally included. generalization of PatchMatch method SuperPatches,...
Superpixel decomposition methods are generally used as a pre-processing step to speed up image processing tasks. They group the pixels of an into homogeneous regions while trying respect existing contours. For all state-of-the-art superpixel methods, trade-off is made between 1) computational time, 2) adherence contours and 3) regularity compactness decomposition. In this paper, we propose fast method compute Superpixels with Contour Adherence using Linear Path (SCALP) in iterative...
In the superpixel literature, comparison of state-of-the-art methods can be biased by nonrobustness some metrics to decomposition aspects, such as scale. Moreover, most recent allow setting a shape regularity parameter, which have substantial impact on measured performances. We introduce an evaluation framework that aims unify process methods. investigate limitations existing and propose evaluate each three core aspects: color homogeneity, respect image objects, regularity. To measure...
This paper deals with the problem of image colorization. A model including total variation regularization is proposed. Our approach colorizes directly three RGB channels, while most existing methods were only focusing on two chrominance channels. By using our able to better preserve color consistency. non convex, but we propose an efficient primal-dual like algorithm compute a local minimizer. Numerical examples illustrate good behavior respect state-of-the-art methods.
Regular decompositions are necessary for most superpixel-based object recognition or tracking applications. So far in the literature, regularity compactness of a superpixel shape is mainly measured by its circularity. In this work, we first demonstrate that such measure not adapted super-pixel evaluation, since it does directly express but circular appearance. Then, propose new metric considers several aspects: convexity, balanced repartition, and contour smoothness. Finally, our robust to...