Résumé: La détection d’objets en mouvement est une étape cruciale pour les systèmes de vidéosurve... more Résumé: La détection d’objets en mouvement est une étape cruciale pour les systèmes de vidéosurveillance. Les mouvements détectés par les algorithmes classiques ne sont pas nécessairement intéressants pour cette application, et le besoin de distinguer les mouvements cohérents des mouvements parasites existe dans la plupart des cas. Dans cet article, nous proposons une nouvelle approche pour faire cette distinction en utilisant l’analyse en composantes principales (ACP), technique issue de l’analyse de données et rarement employée pour ce genre de problèmes. Nous considérons une sous-séquence d’une dizaine de trames, où chaque trame est associée à une dimension de l’espace de représentation des données, dans lequel nous appliquons une ACP afin de pouvoir représenter les données dans un espace de faible dimension dans lequel les points représentant un mouvement cohérent sont proches. Les trames sont ensuite découpées en blocs carrés qui seront projetés dans le nouvel espace de représe...
We present in this paper a new method to track moving objects. It is based on snakes or active co... more We present in this paper a new method to track moving objects. It is based on snakes or active contour models. We are concerned with football game analysis and so the tracked objects are representing football players. The camera is moving too. Our active contour algorithm does not need any preprocessing step contrary to most of the snake-based methods. It is based on classical energies used in active contour algorithms but also on a balloon energy in order to reduce the contour to fit the tracked object. The tracking step does not include any position prediction and is based on a snake initialisation followed by snake deformation. The method implemented is fast enough to consider a real-time framework and has been successfully tested on football game image sequences.
This paper addresses the problem of word spotting in handwritten documents. The method is segment... more This paper addresses the problem of word spotting in handwritten documents. The method is segmentation-free and follows the query-by-string paradigm. In the paper, we focus on the first step of the whole bio-inspired process that is based on two filtering steps, which are a global filtering followed by a more local filtering after a change of observation scale. The contribution of this approach is the use and the generalization of the Haar-Like-Features for the analysis of the document images, inspired from the famous visual perception principle. Different pieces of information are extracted from the whole image before drawing a conclusion, after a process of accumulation of votes. The method is evaluated using the IAM Handwriting Database.
This paper addresses the problem of word spotting in handwritten documents. We propose a coarse-t... more This paper addresses the problem of word spotting in handwritten documents. We propose a coarse-to-fine segmentation free approach. This approach is based on two filtering phases, which are a global filtering followed by a local filtering after changing the observation scale. The contribution of this work is the use and the adaptation of the Haarlike-features in word spotting task for each tested document and the introduction of a new technique permits modelling queries typed by the user. The approach is evaluated using the George Washington manuscripts database. MOTS-CLÉS : Word spotting, caractéristiques pseudo-Haar, analyse de documents
2015 13th International Conference on Document Analysis and Recognition (ICDAR), 2015
In this paper, a Word Spotting model is presented, that is motivated by some characteristics of t... more In this paper, a Word Spotting model is presented, that is motivated by some characteristics of the human visual system. The proposed bio-inspired model works at two different levels. First, a Global Filtering module enables to define several candidate zones. Then, a Refining Filtering module facilitates the selection of good retrieved results. These two modules are based on a process of accumulation of votes resulting from the application of generalized Haar-Like-features. The process does not need the segmentation of documents neither in lines nor in words. The proposed approach is evaluated using the George Washington Database and outperforming state-of-the-art performances.
L'objectif de cette étude est la segmentation du muscle pectoral sur une mammographie. Ce tra... more L'objectif de cette étude est la segmentation du muscle pectoral sur une mammographie. Ce travail a pour but de faciliter le recalage géométrique de paires d'images mammographiques par l'extraction fiable de points de repère. La sémantique de l'image est exploitée pour extraire des points de repère basés sur l'anatomie. La technique d'extraction utilisée est basée sur les contours actifs pour lesquels nous proposons de construire différentes énergies.
Medical Imaging 2013: Computer-Aided Diagnosis, 2013
ABSTRACT This paper aims to detect the evolution between two images representing the same scene. ... more ABSTRACT This paper aims to detect the evolution between two images representing the same scene. The evolution detection problem has many practical applications, especially in medical images. Indeed, the concept of a patient "file" implies the joint analysis of different acquisitions taken at different times, and the detection of significant modifications. The research presented in this paper is carried out within the application context of the development of computer assisted diagnosis (CAD) applied to mammograms. It is performed on already registered pair of images. As the registration is never perfect, we must develop a comparison method sufficiently adapted to detect real small differences between comparable tissues. In many applications, the assessment of similarity used during the registration step is also used for the interpretation step that yields to prompt suspicious regions. In our case registration is assumed to match the spatial coordinates of similar anatomical elements. In this paper, in order to process the medical images at tissue level, the image representation is based on elementary patterns, therefore seeking patterns, not pixels. Besides, as the studied images have low entropy, the decomposed signal is expressed in a parsimonious way. Parsimonious representations are known to help extract the significant structures of a signal, and generate a compact version of the data. This change of representation should allow us to compare the studied images in a short time, thanks to the low weight of the images thus represented, while maintaining a good representativeness. The good precision of our results show the approach efficiency.
Résumé: La détection d’objets en mouvement est une étape cruciale pour les systèmes de vidéosurve... more Résumé: La détection d’objets en mouvement est une étape cruciale pour les systèmes de vidéosurveillance. Les mouvements détectés par les algorithmes classiques ne sont pas nécessairement intéressants pour cette application, et le besoin de distinguer les mouvements cohérents des mouvements parasites existe dans la plupart des cas. Dans cet article, nous proposons une nouvelle approche pour faire cette distinction en utilisant l’analyse en composantes principales (ACP), technique issue de l’analyse de données et rarement employée pour ce genre de problèmes. Nous considérons une sous-séquence d’une dizaine de trames, où chaque trame est associée à une dimension de l’espace de représentation des données, dans lequel nous appliquons une ACP afin de pouvoir représenter les données dans un espace de faible dimension dans lequel les points représentant un mouvement cohérent sont proches. Les trames sont ensuite découpées en blocs carrés qui seront projetés dans le nouvel espace de représe...
We present in this paper a new method to track moving objects. It is based on snakes or active co... more We present in this paper a new method to track moving objects. It is based on snakes or active contour models. We are concerned with football game analysis and so the tracked objects are representing football players. The camera is moving too. Our active contour algorithm does not need any preprocessing step contrary to most of the snake-based methods. It is based on classical energies used in active contour algorithms but also on a balloon energy in order to reduce the contour to fit the tracked object. The tracking step does not include any position prediction and is based on a snake initialisation followed by snake deformation. The method implemented is fast enough to consider a real-time framework and has been successfully tested on football game image sequences.
This paper addresses the problem of word spotting in handwritten documents. The method is segment... more This paper addresses the problem of word spotting in handwritten documents. The method is segmentation-free and follows the query-by-string paradigm. In the paper, we focus on the first step of the whole bio-inspired process that is based on two filtering steps, which are a global filtering followed by a more local filtering after a change of observation scale. The contribution of this approach is the use and the generalization of the Haar-Like-Features for the analysis of the document images, inspired from the famous visual perception principle. Different pieces of information are extracted from the whole image before drawing a conclusion, after a process of accumulation of votes. The method is evaluated using the IAM Handwriting Database.
This paper addresses the problem of word spotting in handwritten documents. We propose a coarse-t... more This paper addresses the problem of word spotting in handwritten documents. We propose a coarse-to-fine segmentation free approach. This approach is based on two filtering phases, which are a global filtering followed by a local filtering after changing the observation scale. The contribution of this work is the use and the adaptation of the Haarlike-features in word spotting task for each tested document and the introduction of a new technique permits modelling queries typed by the user. The approach is evaluated using the George Washington manuscripts database. MOTS-CLÉS : Word spotting, caractéristiques pseudo-Haar, analyse de documents
2015 13th International Conference on Document Analysis and Recognition (ICDAR), 2015
In this paper, a Word Spotting model is presented, that is motivated by some characteristics of t... more In this paper, a Word Spotting model is presented, that is motivated by some characteristics of the human visual system. The proposed bio-inspired model works at two different levels. First, a Global Filtering module enables to define several candidate zones. Then, a Refining Filtering module facilitates the selection of good retrieved results. These two modules are based on a process of accumulation of votes resulting from the application of generalized Haar-Like-features. The process does not need the segmentation of documents neither in lines nor in words. The proposed approach is evaluated using the George Washington Database and outperforming state-of-the-art performances.
L'objectif de cette étude est la segmentation du muscle pectoral sur une mammographie. Ce tra... more L'objectif de cette étude est la segmentation du muscle pectoral sur une mammographie. Ce travail a pour but de faciliter le recalage géométrique de paires d'images mammographiques par l'extraction fiable de points de repère. La sémantique de l'image est exploitée pour extraire des points de repère basés sur l'anatomie. La technique d'extraction utilisée est basée sur les contours actifs pour lesquels nous proposons de construire différentes énergies.
Medical Imaging 2013: Computer-Aided Diagnosis, 2013
ABSTRACT This paper aims to detect the evolution between two images representing the same scene. ... more ABSTRACT This paper aims to detect the evolution between two images representing the same scene. The evolution detection problem has many practical applications, especially in medical images. Indeed, the concept of a patient "file" implies the joint analysis of different acquisitions taken at different times, and the detection of significant modifications. The research presented in this paper is carried out within the application context of the development of computer assisted diagnosis (CAD) applied to mammograms. It is performed on already registered pair of images. As the registration is never perfect, we must develop a comparison method sufficiently adapted to detect real small differences between comparable tissues. In many applications, the assessment of similarity used during the registration step is also used for the interpretation step that yields to prompt suspicious regions. In our case registration is assumed to match the spatial coordinates of similar anatomical elements. In this paper, in order to process the medical images at tissue level, the image representation is based on elementary patterns, therefore seeking patterns, not pixels. Besides, as the studied images have low entropy, the decomposed signal is expressed in a parsimonious way. Parsimonious representations are known to help extract the significant structures of a signal, and generate a compact version of the data. This change of representation should allow us to compare the studied images in a short time, thanks to the low weight of the images thus represented, while maintaining a good representativeness. The good precision of our results show the approach efficiency.
Uploads
Papers by Nicole Vincent