Reducing the amount of keypoints used to index an image is particularly interesting to control processing time and memory usage in real-time document image matching applications, like augmented documents or smartphone applications. This paper benchmarks two keypoint selection methods on a task consisting of reducing keypoint sets extracted from document images, while preserving detection and segmentation accuracy. We first study the different forms of keypoint filtering, and we introduce the use of the CORE selection method on keypoints extracted from document images. Then, we extend a previously published benchmark by including evaluations of the new method, by adding the SURF-BRISK detection/description scheme, and by reporting processing speeds. Evaluations are conducted on the publicly available dataset of ICDAR2015 SmartDOC challenge 1. Finally, we prove that reducing the original keypoint set is always feasible and can be beneficial not only to processing speed but also to accuracy.

The latest developments in sensor design for remote sensing and Earth observation purposes are leading to images always more complex to analyze. Low-level pixel-based processing is becoming unadapted to efficiently handle the wealth of information they contain, and higher levels of abstraction are required. Region-based representations intend to exploit images as collections of regions of interest bearing some semantic meaning, thus easing their interpretation. However, the scale of analysis of the images has to be fixed beforehand, which can be problematic as different applications may not require the same scale of analysis. On the other hand, hierarchical representations are multiscale descriptions of images, as they encompass in their structures all potential regions of interest, organized in a hierarchical manner. Thus, they allow to explore the image at various levels of details and can serve as a single basis for many different further processings. Thanks to its flexibility, the binary partition tree (BPT) representation is one of the most popular hierarchical representations, and has received a lot of attention lately. This article draws a comprehensive review of the most recent works involving BPT representations for various remote sensing data analysis tasks, such as image segmentation and filtering, object detection or hyperspectral classification, and anomaly detection.

Quotient operators have been rarely studied in the context of weighted rational expressions and automaton generation—in spite of the key role played by the quotient of words in formal language theory. To handle both left- and right-quotients we generalize an expansion-based construction of the derived-term (or Antimirov, or equation) automaton and rely on support for a transposition (or reversal) operator. The resulting automata may have spontaneous transitions, which requires different techniques from the usual derived-term constructions.

In digital topology, it is well-known that, in 2D and in 3D, a digital set $X \subseteq Z^n$ is <i>digitally well-composed (DWC)</i>, <i>i.e.</i>, does not contain any critical configuration, if its immersion in the Khalimsky grids $H^n$ is <i>well-composed in the sense of Alexandrov (AWC)</i>, <i>i.e.</i>, its boundary is a disjoint union of discrete $(n-1)$-surfaces. We show that this is still true in $n$-D, $n \geq 2$, which is of prime importance since today 4D signals are more and more frequent.

De nombreuses applications biomedicales impliquent l’analyse de séquences pour la caractérisation du mouvement. Dans cet article, nous considerons des séquences 2D+t où un mouvement particulier (par exemple un flux sanguin) est associé à une zone spécifique de l’image 2D (par exemple une artère). Mais de nombreux mouvements peuvent co-exister dans les séquences (par exemple, il peut y avoir plusieurs vaisseaux sanguins presents, chacun avec leur flux spécifique). La caractérisation de ce type de mouvement implique d’abord de trouver les zones où le mouvement est présent, puis d’analyser ces mouvements : vitesse, régularité, fréquence, etc. Dans cet article, nous proposons une méthode appropriée pour détecter et caractériser simultanément les zones où le mouvement est présent dans une séquence. Nous pouvons ensuite classer ce mouvement en zones cohérentes en utilisant un apprentissage non supervisé et produire des métriques directement utilisables pour diverses applications. Nous illustrons et validons cette même méthode sur l’analyse du flux sanguin chez l’embryon de poisson.

L’imagerie par résonance magnétique (IRM) du cerveau est utilisée sur les nouveau-nés pour évaluer l’évolution du cerveau et diagnostiquer des maladies neurologiques. Ces examens nécessitent souvent une analyse quantitative des différents tissus du cerveau, de sorte qu’avoir une segmentation précise est essentiel. Dans cet article, nous proposons une méthode automatique rapide de segmentation en différents tissus des images IRM 3D de cerveaux de nouveau-nés ; elle utilise un réseau de neurones totalement convolutif (FCN) et du transfert d’apprentissage. Par rapport aux approches similaires qui reposent soit sur des patchs 2D ou 3D, soit sur des FCN totalement 3D, notre méthode est beaucoup plus rapide : elle ne prend que quelques secondes, et une seule modalité (T2) est nécessaire. Afin de prendre les informations 3D en compte, trois coupes 2D successives sont empilées pour former une image 2D en couleurs, dont l’ensemble sur tout le volume sert d’entrée à un FCN, pré-entraîné sur ImageNet pour la classification d’images naturelles. Nos expériences sur un ensemble de données de référence montrent que notre méthode obtient des résultats du niveau de ceux de l’état de l’art.

Brain magnetic resonance imaging (MRI) is widely used to assess brain developments in neonates and to diagnose a wide range of neurological diseases in adults. Such studies are usually based on quantitative analysis of different brain tissues, so it is essential to be able to classify them accurately. In this paper, we propose a fast automatic method that segments 3D brain MR images into different tissues using fully convolutional network (FCN) and transfer learning. As compared to existing deep learning-based approaches that rely either on 2D patches or on fully 3D FCN, our method is way much faster: it only takes a few seconds, and only a single modality (T1 or T2) is required. In order to take the 3D information into account, all 3 successive 2D slices are stacked to form a set of 2D color images, which serve as input for the FCN pre-trained on ImageNet for natural image classification. To the best of our knowledge, this is the first method that applies transfer learning to segment both neonatal and adult brain 3D MR images. Our experiments on two public datasets show that our method achieves state-of-the-art results.

Over the last decade, parallel SAT solving has been widely studied from both theoretical and practical aspects. There are now numerous solvers that dier by parallelization strategies, programming languages, concurrent programming, involved libraries, etc. Hence, comparing the eciency of the theoretical approaches is a challenging task. Moreover, the introduction of a new approach needs either a deep understanding of the existing solvers, or to start from scratch the implementation of a new tool. We present PaInleSS: a framework to build parallel SAT solvers for many-core environments. Thanks to its genericity and modularity, it provides the implementation of basics for parallel SAT solving like clause exchanges, Portfolio and Divide-and-Conquer strategies. It also enables users to easily create their own parallel solvers based on new strategies. Our experiments show that our framework compares well with some of the best state-of-the-art solvers.

It is now possible to collect hyperspectral video sequences at a near real-time frame rate. The wealth of spectral, spatial and temporal information of those sequences is appealing for various applications, but classical video processing techniques must be adapted to handle the high dimensionality and huge size of the data to process. In this article, we introduce a novel method based on the hierarchical analysis of hyperspectral video sequences to perform object tracking. This latter operation is tackled as a sequential object detection process, conducted on the hierarchical representation of the hyperspectral video frames. We apply the proposed methodology to the chemical gas plume tracking scenario and compare its performances with state-of-the-art methods, for two real hyperspectral video sequences, and show that the proposed approach performs at least equally well.

In the automata theoretic approach to explicit state LTL model checking, the synchronized product of the model and an automaton that represents the negated formula is checked for emptiness. In practice, a (transition-based generalized) Büchi automaton (TGBA) is used for this procedure.This paper investigates whether using a more general form of acceptance, namely transition-based generalized Rabin automata (TGRAs), improves the model checking procedure. TGRAs can have significantly fewer states than TGBAs, however the corresponding emptiness checking procedure is more involved. With recent advances in probabilistic model checking and LTL to TGRA translators, it is only natural to ask whether checking a TGRA directly is more advantageous in practice.We designed a multi-core TGRA checking algorithm and performed experiments on a subset of the models and formulas from the 2015 Model Checking Contest. We observed that our algorithm can be used to replace a TGBA checking algorithm without losing performance. In general, we found little to no improvement by checking TGRAs directly.