Texture discrimination was the second more important task studied after colour perception and cha... more Texture discrimination was the second more important task studied after colour perception and characterization. Nevertheless, few works explore the colour extension of these works and none for vectorial processing of this important visual information. In this work we propose a novel and vector processing for colour texture characterization, the color contrast occurrence matrix C 2 O. This new texture feature is based on the colour difference assessment. To be link to the human perception, the colour difference is expressed using a perceptual distance expressed in CIELab and two angles characterizing the chromaticity and darker or lighter direction. Through this new attribute, we analyze the stability to changes in illumination, viewpoint and spectrum of the light source in front of different texture image databases. Thanks to our construction, we avoid the main limit of existing texture features requiring an initial colour quantization or a binarization inside the texture construction. Keeping the small local contrast, we obtain a more accurate texture feature description explaining the obtained results. Then we carry out the construction of a features vector by occurrence quantization, keeping the initial ideas of Julesz, Haralick and Ojala, for the classification purposes. The results show best correct classification percentages in databases that with important spatio-chromatic complexity as ALOT.
Abstract A hyperspectral sensor is able to acquire the physical and optical properties of surface... more Abstract A hyperspectral sensor is able to acquire the physical and optical properties of surfaces. On the other hand, applications in climate research, ecology, cultural heritage, and industry require metrological solutions for diagnosis and control quality. Thus there is a direct relationship between the physical properties of materials and their processing results obtained from spectral measurements. Two types of measurements are considered. First, each material of interest can be identified or modeled through a single spectrum. The second type considers that a material does not present a uniform response. Therefore its identification can be done through the distribution of its spectra. In order to preserve the metrological properties of the acquired measurements while also solving the “curse of dimensionality” problem, these considerations are expressed in the domain of spectral differences. Protocols to assess the metrological properties of difference functions are provided, along with applications in remote sensing data analysis.
A lot of image processing tasks require key-point detection. If grey-level approach are numerous,... more A lot of image processing tasks require key-point detection. If grey-level approach are numerous, colour and hyper-spectral ones are scarce. In this paper, we propose a generic key-point detection for colour, multi and hyper-spectral images. A new synthetic database is created to compare key-point detection approaches. Our method improves detection when the image complexity increases.
With the emergence of huge volumes of high-resolution Hyperspectral Images (HSI)produced by diffe... more With the emergence of huge volumes of high-resolution Hyperspectral Images (HSI)produced by different types of imaging sensors, analyzing and retrieving these images requireeffective image description and quantification techniques. Compared to remote sensing RGB images,HSI data contain hundreds of spectral bands (varying from the visible to the infrared ranges) allowingprofile materials and organisms that only hyperspectral sensors can provide. In this article, we studythe importance of spectral sensitivity functions in constructing discriminative representation ofhyperspectral images. The main goal of such representation is to improve image content recognitionby focusing the processing on only the most relevant spectral channels. The underlying hypothesisis that for a given category, the content of each image is better extracted through a specific set ofspectral sensitivity functions. Those spectral sensitivity functions are evaluated in a Content-BasedImage Retrieval (CBIR) framew...
Spectral Information Divergence (SID) was identified as the most efficient spectral similarity me... more Spectral Information Divergence (SID) was identified as the most efficient spectral similarity measure. However, we show that divergence are not adapted to direct use on spectra. Following an idea proposed by Nidamanuri, we construct a spectral pseudo-divergence based on the Kullback-Leibler divergence. This pseudo-divergence is composed of two parts: a shape and an intensity similarity measure. Consequently, bidimensional representation of spectral differences are constructed to display the histograms of similarity between a spectral reference and the spectra from a data-set or an hyperspectral image. We prove the efficiency of the spectral similarity measure and of the bidimensional histogram of spectral differences on artificial and Cultural Heritage spectral images.
In this paper, we explore an original way to compute texture features for color images in a vecto... more In this paper, we explore an original way to compute texture features for color images in a vector process. Using a dedicated approach for color ordering, we produce a complete framework for color mathematical morphology adapted to human visual system characteristics. Then, morphological multiscale texture features are defined. To understand the texture feature behavior, we present the feature response to basic images variations. Finally, we compare the texture feature performance in front of a classical classification task using Outex database.
DESCRIPTION Two fully vector texture features are compared to classical texture features for colo... more DESCRIPTION Two fully vector texture features are compared to classical texture features for colour images classification. Behind this comparison, the asked question is about theb required complexity for colour texture discrimination. To explore it, we start from a classical work on colour texture classification from Arvis. We processed Cooccurrence and Run-length matrix in a multi-components way and with two adapted grey-level approaches (from the colour indexed equivalent image and from the Intensity image completed by colour moments). Then we compare the obtained results to more advanced texture features, computed in a vector way from the colour image, and from the intensity image. These features are the morphological covariance and the fractal signature. Results are computed for outex and vistex database.
For a complex writting as egyptian hieroglyphs, combining the works done in hierarchical modeliza... more For a complex writting as egyptian hieroglyphs, combining the works done in hierarchical modelizations and fuzzy grammar definitions seems natural. This paper introduce the hierarchical-fuzzy-attributed graph (FHAG), extended from fuzzy-attributed graph, which modelize attributes by fuzzy-tree grammar. We give a formal definition of FHAGs and explain the building process. Some results are given with a recognition system based on single models comparisons.
2013 IEEE 24th Annual International Symposium on Personal, Indoor, and Mobile Radio Communications (PIMRC), 2013
ABSTRACT In Wireless Sensor Networks, where the communicating nodes are not mobile, the channel v... more ABSTRACT In Wireless Sensor Networks, where the communicating nodes are not mobile, the channel variations are mainly temporal and caused by the random displacement of human beings in the environment. In this paper, we propose to model these kinds of variations for transmissions at 433 and 868 MHz according to the number of persons in motion. The proposed approach is independent of the environment geometry and the spatial variability of the channel, but is complementary to the models that take into account these variations. It relies on a narrowband measurements campaign in order to predict the channel behaviour with a statistical identification method. Finally, a temporal modelling compliant to building monitoring applications is used to assess the impact of people motion on the transmission quality through the computation of the associated Bit Error Rates.
In a sustainable development context, the monitoring systems are essential to study the building ... more In a sustainable development context, the monitoring systems are essential to study the building energy performances. With the recent technology advances, these systems can be based on wireless sensor networks, where the energy efficiency is the main design challenge. To this end, most of the studies focus on low power Medium Access Control (MAC) protocols to reduce the overall energy consumption of a network. Nevertheless, the performances assessment of these protocols is generally not performed in a realistic way, and does not take into account the performances of the other layers of the OSI model. In this paper, we propose a cross-layer methodology to assess the real performances of a MAC protocol by taking into account the traffic volume, the synchronization losses and more particularly the physical layer performances through a Bit Error Rate (BER) criterion. The simulation results demonstrate clearly the physical layer impact on a sensor lifetime. Finally, the proposal of an energy efficient MAC protocol for a wireless sensor network dedicated to an application of building monitoring is proposed.
ABSTRACT This paper describes how a query system can exploit the basic knowledge by employing sem... more ABSTRACT This paper describes how a query system can exploit the basic knowledge by employing semi-automatic relevance feedback to refine queries and runtimes. For general databases, it is often useless to call complex attributes, because we have not sufficient information about images in the database. Moreover, these images can be topologically very different from one to each other and an attribute that is powerful for a database category may be very powerless for the other categories. The idea is to use very simple features, such as color histogram, correlograms, Color Coherence Vectors (CCV), to fill out the signature vector. Then, a number of mixture vectors is prepared depending on the number of very distinctive categories in the database. Knowing that a mixture vector is a vector containing the weight of each attribute that will be used to compute a similarity distance. We post a query in the database using successively all the mixture vectors defined previously. We retain then the N first images for each vector in order to make a mapping using the following information: Is image I present in several mixture vectors results? What is its rank in the results? These informations allow us to switch the system on an unsupervised relevance feedback or user's feedback (supervised feedback).
In this article we try to make different kinds of information cooperate in a characters recogniti... more In this article we try to make different kinds of information cooperate in a characters recognition system addressing old Greek and Egyptians documents. We first use a statistical approach based on classical shape descriptors (Zernike, Fourier). Then we use a structural classification method with an attributed graph description of characters and a random graph modeling of classes. The hypothesis, that structural methods bring topological information that statistical methods do not, is validated on Greek characters. A cooperation with a chain of classifiers based on reject management is then proposed. Due to computation cost, the goal of such a chain is to use the structural approach only if the statistical one fails.
In this paper, we explore an original way to compute texture features for color images in a vecto... more In this paper, we explore an original way to compute texture features for color images in a vector process. To do it, we used a dedicated approach for color mathematical morphology using distance function. We show in this paper the scientific construction of morphological spectra and preliminary results using Outex database.
Proceedings. First International Symposium on 3D Data Processing Visualization and Transmission
The aim of this study is to present research made in a 3D scanner conception of a human foot. The... more The aim of this study is to present research made in a 3D scanner conception of a human foot. The first step of foot reconstruction is treated. The cloud number points is increased thanks to a combination of several methods. The first method for the first points extraction is explained. The relative length, height and width of the object are obtained in order to construct the parallelepiped including the foot. All parallelepiped points are virtually assigned to foot points. Each image is projected in the volume in the different planes of the views in order to eliminate the exterior points. From the volume of all these points only the points of the surface are retained. The points of the cloud should be located on or outside the surface of the foot. This reconstruction must be ameliorated by an active triangulation. We use the tools of matching points of stereovision in the image profiles.
Résumé En indexation d’images, la plupart des descripteurs développés sont basés sur des statist... more Résumé En indexation d’images, la plupart des descripteurs développés sont basés sur des statistiques globales des images. Aujourd’hui, des méthodes essayant d’aller plus loin dans la représentation apparaissent. Sans aller jusqu ’àune description sémantique, elles représentent les images plus finement en utilisant des statistiques localisées ainsi que les interactions entre celles-ci. Dans la plupart des cas, des structures de graphes sont
Texture discrimination was the second more important task studied after colour perception and cha... more Texture discrimination was the second more important task studied after colour perception and characterization. Nevertheless, few works explore the colour extension of these works and none for vectorial processing of this important visual information. In this work we propose a novel and vector processing for colour texture characterization, the color contrast occurrence matrix C 2 O. This new texture feature is based on the colour difference assessment. To be link to the human perception, the colour difference is expressed using a perceptual distance expressed in CIELab and two angles characterizing the chromaticity and darker or lighter direction. Through this new attribute, we analyze the stability to changes in illumination, viewpoint and spectrum of the light source in front of different texture image databases. Thanks to our construction, we avoid the main limit of existing texture features requiring an initial colour quantization or a binarization inside the texture construction. Keeping the small local contrast, we obtain a more accurate texture feature description explaining the obtained results. Then we carry out the construction of a features vector by occurrence quantization, keeping the initial ideas of Julesz, Haralick and Ojala, for the classification purposes. The results show best correct classification percentages in databases that with important spatio-chromatic complexity as ALOT.
Abstract A hyperspectral sensor is able to acquire the physical and optical properties of surface... more Abstract A hyperspectral sensor is able to acquire the physical and optical properties of surfaces. On the other hand, applications in climate research, ecology, cultural heritage, and industry require metrological solutions for diagnosis and control quality. Thus there is a direct relationship between the physical properties of materials and their processing results obtained from spectral measurements. Two types of measurements are considered. First, each material of interest can be identified or modeled through a single spectrum. The second type considers that a material does not present a uniform response. Therefore its identification can be done through the distribution of its spectra. In order to preserve the metrological properties of the acquired measurements while also solving the “curse of dimensionality” problem, these considerations are expressed in the domain of spectral differences. Protocols to assess the metrological properties of difference functions are provided, along with applications in remote sensing data analysis.
A lot of image processing tasks require key-point detection. If grey-level approach are numerous,... more A lot of image processing tasks require key-point detection. If grey-level approach are numerous, colour and hyper-spectral ones are scarce. In this paper, we propose a generic key-point detection for colour, multi and hyper-spectral images. A new synthetic database is created to compare key-point detection approaches. Our method improves detection when the image complexity increases.
With the emergence of huge volumes of high-resolution Hyperspectral Images (HSI)produced by diffe... more With the emergence of huge volumes of high-resolution Hyperspectral Images (HSI)produced by different types of imaging sensors, analyzing and retrieving these images requireeffective image description and quantification techniques. Compared to remote sensing RGB images,HSI data contain hundreds of spectral bands (varying from the visible to the infrared ranges) allowingprofile materials and organisms that only hyperspectral sensors can provide. In this article, we studythe importance of spectral sensitivity functions in constructing discriminative representation ofhyperspectral images. The main goal of such representation is to improve image content recognitionby focusing the processing on only the most relevant spectral channels. The underlying hypothesisis that for a given category, the content of each image is better extracted through a specific set ofspectral sensitivity functions. Those spectral sensitivity functions are evaluated in a Content-BasedImage Retrieval (CBIR) framew...
Spectral Information Divergence (SID) was identified as the most efficient spectral similarity me... more Spectral Information Divergence (SID) was identified as the most efficient spectral similarity measure. However, we show that divergence are not adapted to direct use on spectra. Following an idea proposed by Nidamanuri, we construct a spectral pseudo-divergence based on the Kullback-Leibler divergence. This pseudo-divergence is composed of two parts: a shape and an intensity similarity measure. Consequently, bidimensional representation of spectral differences are constructed to display the histograms of similarity between a spectral reference and the spectra from a data-set or an hyperspectral image. We prove the efficiency of the spectral similarity measure and of the bidimensional histogram of spectral differences on artificial and Cultural Heritage spectral images.
In this paper, we explore an original way to compute texture features for color images in a vecto... more In this paper, we explore an original way to compute texture features for color images in a vector process. Using a dedicated approach for color ordering, we produce a complete framework for color mathematical morphology adapted to human visual system characteristics. Then, morphological multiscale texture features are defined. To understand the texture feature behavior, we present the feature response to basic images variations. Finally, we compare the texture feature performance in front of a classical classification task using Outex database.
DESCRIPTION Two fully vector texture features are compared to classical texture features for colo... more DESCRIPTION Two fully vector texture features are compared to classical texture features for colour images classification. Behind this comparison, the asked question is about theb required complexity for colour texture discrimination. To explore it, we start from a classical work on colour texture classification from Arvis. We processed Cooccurrence and Run-length matrix in a multi-components way and with two adapted grey-level approaches (from the colour indexed equivalent image and from the Intensity image completed by colour moments). Then we compare the obtained results to more advanced texture features, computed in a vector way from the colour image, and from the intensity image. These features are the morphological covariance and the fractal signature. Results are computed for outex and vistex database.
For a complex writting as egyptian hieroglyphs, combining the works done in hierarchical modeliza... more For a complex writting as egyptian hieroglyphs, combining the works done in hierarchical modelizations and fuzzy grammar definitions seems natural. This paper introduce the hierarchical-fuzzy-attributed graph (FHAG), extended from fuzzy-attributed graph, which modelize attributes by fuzzy-tree grammar. We give a formal definition of FHAGs and explain the building process. Some results are given with a recognition system based on single models comparisons.
2013 IEEE 24th Annual International Symposium on Personal, Indoor, and Mobile Radio Communications (PIMRC), 2013
ABSTRACT In Wireless Sensor Networks, where the communicating nodes are not mobile, the channel v... more ABSTRACT In Wireless Sensor Networks, where the communicating nodes are not mobile, the channel variations are mainly temporal and caused by the random displacement of human beings in the environment. In this paper, we propose to model these kinds of variations for transmissions at 433 and 868 MHz according to the number of persons in motion. The proposed approach is independent of the environment geometry and the spatial variability of the channel, but is complementary to the models that take into account these variations. It relies on a narrowband measurements campaign in order to predict the channel behaviour with a statistical identification method. Finally, a temporal modelling compliant to building monitoring applications is used to assess the impact of people motion on the transmission quality through the computation of the associated Bit Error Rates.
In a sustainable development context, the monitoring systems are essential to study the building ... more In a sustainable development context, the monitoring systems are essential to study the building energy performances. With the recent technology advances, these systems can be based on wireless sensor networks, where the energy efficiency is the main design challenge. To this end, most of the studies focus on low power Medium Access Control (MAC) protocols to reduce the overall energy consumption of a network. Nevertheless, the performances assessment of these protocols is generally not performed in a realistic way, and does not take into account the performances of the other layers of the OSI model. In this paper, we propose a cross-layer methodology to assess the real performances of a MAC protocol by taking into account the traffic volume, the synchronization losses and more particularly the physical layer performances through a Bit Error Rate (BER) criterion. The simulation results demonstrate clearly the physical layer impact on a sensor lifetime. Finally, the proposal of an energy efficient MAC protocol for a wireless sensor network dedicated to an application of building monitoring is proposed.
ABSTRACT This paper describes how a query system can exploit the basic knowledge by employing sem... more ABSTRACT This paper describes how a query system can exploit the basic knowledge by employing semi-automatic relevance feedback to refine queries and runtimes. For general databases, it is often useless to call complex attributes, because we have not sufficient information about images in the database. Moreover, these images can be topologically very different from one to each other and an attribute that is powerful for a database category may be very powerless for the other categories. The idea is to use very simple features, such as color histogram, correlograms, Color Coherence Vectors (CCV), to fill out the signature vector. Then, a number of mixture vectors is prepared depending on the number of very distinctive categories in the database. Knowing that a mixture vector is a vector containing the weight of each attribute that will be used to compute a similarity distance. We post a query in the database using successively all the mixture vectors defined previously. We retain then the N first images for each vector in order to make a mapping using the following information: Is image I present in several mixture vectors results? What is its rank in the results? These informations allow us to switch the system on an unsupervised relevance feedback or user's feedback (supervised feedback).
In this article we try to make different kinds of information cooperate in a characters recogniti... more In this article we try to make different kinds of information cooperate in a characters recognition system addressing old Greek and Egyptians documents. We first use a statistical approach based on classical shape descriptors (Zernike, Fourier). Then we use a structural classification method with an attributed graph description of characters and a random graph modeling of classes. The hypothesis, that structural methods bring topological information that statistical methods do not, is validated on Greek characters. A cooperation with a chain of classifiers based on reject management is then proposed. Due to computation cost, the goal of such a chain is to use the structural approach only if the statistical one fails.
In this paper, we explore an original way to compute texture features for color images in a vecto... more In this paper, we explore an original way to compute texture features for color images in a vector process. To do it, we used a dedicated approach for color mathematical morphology using distance function. We show in this paper the scientific construction of morphological spectra and preliminary results using Outex database.
Proceedings. First International Symposium on 3D Data Processing Visualization and Transmission
The aim of this study is to present research made in a 3D scanner conception of a human foot. The... more The aim of this study is to present research made in a 3D scanner conception of a human foot. The first step of foot reconstruction is treated. The cloud number points is increased thanks to a combination of several methods. The first method for the first points extraction is explained. The relative length, height and width of the object are obtained in order to construct the parallelepiped including the foot. All parallelepiped points are virtually assigned to foot points. Each image is projected in the volume in the different planes of the views in order to eliminate the exterior points. From the volume of all these points only the points of the surface are retained. The points of the cloud should be located on or outside the surface of the foot. This reconstruction must be ameliorated by an active triangulation. We use the tools of matching points of stereovision in the image profiles.
Résumé En indexation d’images, la plupart des descripteurs développés sont basés sur des statist... more Résumé En indexation d’images, la plupart des descripteurs développés sont basés sur des statistiques globales des images. Aujourd’hui, des méthodes essayant d’aller plus loin dans la représentation apparaissent. Sans aller jusqu ’àune description sémantique, elles représentent les images plus finement en utilisant des statistiques localisées ainsi que les interactions entre celles-ci. Dans la plupart des cas, des structures de graphes sont
Uploads
Papers by Noël Richard