The Experts below are selected from a list of 18882 Experts worldwide ranked by ideXlab platform
Frédéric Dufaux - One of the best experts on this subject based on the ideXlab platform.
-
Deep Tone Mapping Operator for High Dynamic Range Images
IEEE Transactions on Image Processing, 2019Co-Authors: Aakanksha Rana, Praveer Singh, Giuseppe Valenzise, Frédéric Dufaux, Nikos Komodakis, Aljosa SmolicAbstract:A computationally fast tone Mapping Operator (TMO) that can quickly adapt to a wide spectrum of high dynamic range (HDR) content is quintessential for visualization on varied low dynamic range (LDR) output devices such as movie screens or standard displays. Existing TMOs can successfully tone-map only a limited number of HDR content and require an extensive parameter tuning to yield the best subjective-quality tone-mapped output. In this paper, we address this problem by proposing a fast, parameter-free and scene-adaptable deep tone Mapping Operator (DeepTMO) that yields a high-resolution and high-subjective quality tone mapped output. Based on conditional generative adversarial network (cGAN), DeepTMO not only learns to adapt to vast scenic-content (e.g., outdoor, indoor, human, structures, etc.) but also tackles the HDR related scene-specific challenges such as contrast and brightness, while preserving the fine-grained details. We explore 4 possible combinations of Generator-Discriminator architectural designs to specifically address some prominent issues in HDR related deep-learning frameworks like blurring, tiling patterns and saturation artifacts. By exploring different influences of scales, loss-functions and normalization layers under a cGAN setting, we conclude with adopting a multi-scale model for our task. To further leverage on the large-scale availability of unlabeled HDR data, we train our network by generating targets using an objective HDR quality metric, namely Tone Mapping Image Quality Index (TMQI). We demonstrate results both quantitatively and qualitatively, and showcase that our DeepTMO generates high-resolution, high-quality output images over a large spectrum of real-world scenes. Finally, we evaluate the perceived quality of our results by conducting a pair-wise subjective study which confirms the versatility of our method.
-
learning based tone Mapping Operator for efficient image matching
IEEE Transactions on Multimedia, 2019Co-Authors: Aakanksha Rana, Giuseppe Valenzise, Frédéric DufauxAbstract:In this paper, we propose a new framework to optimally tone map the high dynamic range (HDR) content for image matching under drastic illumination variations. Since tone Mapping Operators (TMO) have traditionally been used for displaying HDR scenes, their design is suboptimal when used for computer vision tasks, such as image matching. We address this suboptimality by proposing a two-step framework, consisting of: first, a luminance-invariant guidance model based on a support vector regressor (SVR) to optimally adapt the tone Mapping function for image matching; and second, an energy maximization model to generate appropriate training samples for learning the SVR. At each step, we collectively address both stages of keypoint detection and descriptor extraction in the feature matching framework. By locally altering the intrinsic characteristics of the tone Mapping function, the learned guidance model facilitates the extraction of local invariant features in the presence of illumination variations. We demonstrate that the proposed TMO significantly outperforms perceptually driven state-of-the-art TMOs on a dataset of HDR scenes characterized by challenging lighting variations, such as day/night transitions.
-
spatio temporal constrained tone Mapping Operator for hdr video compression
Journal of Visual Communication and Image Representation, 2018Co-Authors: Giuseppe Valenzise, Cagri Ozcinar, Paul Lauga, Frédéric DufauxAbstract:Abstract With the growing popularity of high dynamic range (HDR) imaging, efficient compression techniques are demanded, as HDR video entails typically higher raw data rate than traditional video. For this purpose, we introduce a hybrid spatially and temporally constrained content-adaptive tone Mapping Operator (TMO) to convert the input HDR video into a tone mapped video sequence, which is then encoded using the high efficiency video coding (HEVC) standard. The proposed TMO simultaneously exploits intra-frame spatial redundancies and preserves inter-frame temporal coherence of the tone mapped video sequence. Extensive experimental results show that the developed spatio-temporal TMO (ST-TMO) solution yields higher coding performance than existing frame-by-frame TMO’s, and compares favorably with state-of-the-art methods based on a fixed transfer function.
-
Learning-Based Tone Mapping Operator for Image Matching
2017Co-Authors: Aakanksha Rana, Giuseppe Valenzise, Frédéric DufauxAbstract:In this paper, we propose a new framework to optimally tone-map a high dynamic range (HDR) content for image matching under drastic illumination variations. This task is of fundamental importance for many computer vision applications. To design such a framework, we build a luminance invariant guidance model using a Support Vector Regressor (SVR) and learn it to facilitate the extraction of invariant descriptors from scenes subject to wide variety of appearance changes such as day/night transition. To this end, we initially generate appropriate training samples using a simple similarity-maximization mechanism. We then employ the learned model to predict optimal modulation maps that help to locally alter the intrinsic characteristics (such as shape, size) of the tone Mapping function. We evaluate the proposed model performance in terms of matching score and mean average precision rate using state-of-the-art descriptor extraction schemes. We demonstrate that our tone Mapping framework significantly outperforms the existing perceptually-driven state-of-the-art TMOs on the benchmark datasets.
-
HDR Video Coding based on a Temporally Constrained Tone Mapping Operator,
2016Co-Authors: Cagri Ozcinar, Giuseppe Valenzise, Paul Lauga, Frédéric DufauxAbstract:Given its potential for more realistic rendering and enhanced user experience, High Dynamic Range (HDR) imaging is raising a lot of interest both in industry and academia. In this context, efficient representation and coding techniques are needed, as HDR video entails significantly higher raw data rate. In this paper, we present a temporally constrained contentadaptive Tone Mapping Operator (TMO) in order to convert the input HDR video into a reduced bit depth video sequence which is then encoded using High Efficiency Video Coding (HEVC). As the proposed TMO simultaneously takes into account the statistical characteristics of the input frame while better preserving temporal coherence of the tone mapped video sequence, it leads to improved coding efficiency. Experimental results show that the proposed technique compares favorably with existing methods in terms of rate-distortion when using the HDR-VDP-2.2.1 quality metric.
Vuik Cornelis - One of the best experts on this subject based on the ideXlab platform.
-
An IGA Framework for PDE-Based Planar Parameterization on Convex Multipatch Domains
'Springer Science and Business Media LLC', 2021Co-Authors: Hinz J.p., Vuik Cornelis, Möller M., Vuik C., Van Brummelen H., Verhoosel C., Simeon B.Abstract:The first step towards applying isogeometric analysis techniques to solve PDE problems on a given domain consists in generating an analysis-suitable Mapping Operator between parametric and physical domains with one or several patches from no more than a description of the boundary contours of the physical domain. A subclass of the multitude of the available parameterization algorithms are those based on the principles of Elliptic Grid Generation (EGG) which, in their most basic form, attempt to approximate a Mapping Operator whose inverse is composed of harmonic functions. The main challenge lies in finding a formulation of the problem that is suitable for a computational approach and a common strategy is to approximate the Mapping Operator by means of solving a PDE-problem. PDE-based EGG is well-established in classical meshing and first generalization attempts to spline-based descriptions (as is mandatory in IgA) have been made. Unfortunately, all of the practically viable PDE-based approaches impose certain requirements on the employed spline-basis, in particular global C≥1-continuity.This paper discusses an EGG-algorithm for the generation of planar parameterizations with locally reduced smoothness (i.e., with support for locally only C0-continuous bases). A major use case of the proposed algorithm is that of multipatch parameterizations, made possible by the support of C0-continuities. This paper proposes a specially-taylored solution algorithm that exploits many characteristics of the PDE-problem and is suitable for large-scale applications. It is discussed for the single-patch case before generalizing its concepts to multipatch settings. This paper is concluded with three numerical experiments and a discussion of the results
-
An IGA Framework for PDE-Based Planar Parameterization on Convex Multipatch Domains
'Springer Science and Business Media LLC', 2021Co-Authors: Hinz J.p., Möller M., Vuik CornelisAbstract:The first step towards applying isogeometric analysis techniques to solve PDE problems on a given domain consists in generating an analysis-suitable Mapping Operator between parametric and physical domains with one or several patches from no more than a description of the boundary contours of the physical domain. A subclass of the multitude of the available parameterization algorithms are those based on the principles of Elliptic Grid Generation (EGG) which, in their most basic form, attempt to approximate a Mapping Operator whose inverse is composed of harmonic functions. The main challenge lies in finding a formulation of the problem that is suitable for a computational approach and a common strategy is to approximate the Mapping Operator by means of solving a PDE-problem. PDE-based EGG is well-established in classical meshing and first generalization attempts to spline-based descriptions (as is mandatory in IgA) have been made. Unfortunately, all of the practically viable PDE-based approaches impose certain requirements on the employed spline-basis, in particular global C≥1-continuity.This paper discusses an EGG-algorithm for the generation of planar parameterizations with locally reduced smoothness (i.e., with support for locally only C0-continuous bases). A major use case of the proposed algorithm is that of multipatch parameterizations, made possible by the support of C0-continuities. This paper proposes a specially-taylored solution algorithm that exploits many characteristics of the PDE-problem and is suitable for large-scale applications. It is discussed for the single-patch case before generalizing its concepts to multipatch settings. This paper is concluded with three numerical experiments and a discussion of the results.Green Open Access added to TU Delft Institutional Repository ‘You share, we take care!’ – Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.Numerical Analysi
-
Spline-based meshing techniques for industrial applications
International Centre for Numerical Methods in Engineering CIMNE, 2020Co-Authors: Hinz J.p., Mölle M., Vuik CornelisAbstract:Isogeometric Analysis (IgA) has become an accepted framework for the mod-elling, simulation and optimization (MSO) of engineering processes. However, the fully automatized generation of analysis-suitable parameterizations of geometries as they arise in practical workflows is still a challenging task, which often requires application-specific parameterization approaches. In this article we present a practical approach [6] based on the principles of Elliptic Grid Generation (EGG) for the efficient on-demand generation of analysis-suitable spline-based parameterizations. Starting from a (point cloud) description of the boundary provided by the existing MSO-pipeline, an inverse nonlinear Poisson-type problem is solved to obtain a folding-free (planar) parameterization of the entire domain. The non-linearity is efficiently treated with a globalized hierarchical Newton approach. Automatized boundary contour reparameterization techniques are employed to improve the parametric properties from a numerical viewpoint, such as orthogonal isolines and equally-sized cells. The use of curved instead of straight-sided elements allows us to arrive at an accurate description of the target domain with fewer elements and thus potentially lower computational effort. Numerical experiments with screw-compressor geometries demonstrate that the proposed algorithm reliably produces high-quality parameterizations typically within 3−4 Newton-iterations, even in the presence of extreme aspect-ratios. This makes it particularly attractive for the on-demand application within an automatized industrial MSO-pipeline. To support the demands of modern high-performance computing hardware, only a moderate number of sufficiently large and structured patches is generated which can be mapped one-by-one to the different devices (CPUs/GPUs) with only little communication overhead. Topology changes are avoided in time-dependent and shape-optimization settings. Finally, the spline-based geometry description can be transformed into a classical mesh by performing a large number of function evaluations in the Mapping Operator. Its continuous nature allows for feature-based (structured and unstructured) refinement and arbitrary element densities without increasing the complexity of the meshing process.Green Open Access added to TU Delft Institutional Repository ‘You share, we take care!’ – Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.Numerical Analysi
-
Spline-based meshing techniques for industrial applications
International Centre for Numerical Methods in Engineering CIMNE, 2020Co-Authors: Hinz J.p., Vuik Cornelis, Möller M., Owen Roger, De Borst Rene, Reese Jason, Pearce ChrisAbstract:Isogeometric Analysis (IgA) has become an accepted framework for the mod-elling, simulation and optimization (MSO) of engineering processes. However, the fully automatized generation of analysis-suitable parameterizations of geometries as they arise in practical workflows is still a challenging task, which often requires application-specific parameterization approaches. In this article we present a practical approach [6] based on the principles of Elliptic Grid Generation (EGG) for the efficient on-demand generation of analysis-suitable spline-based parameterizations. Starting from a (point cloud) description of the boundary provided by the existing MSO-pipeline, an inverse nonlinear Poisson-type problem is solved to obtain a folding-free (planar) parameterization of the entire domain. The non-linearity is efficiently treated with a globalized hierarchical Newton approach. Automatized boundary contour reparameterization techniques are employed to improve the parametric properties from a numerical viewpoint, such as orthogonal isolines and equally-sized cells. The use of curved instead of straight-sided elements allows us to arrive at an accurate description of the target domain with fewer elements and thus potentially lower computational effort. Numerical experiments with screw-compressor geometries demonstrate that the proposed algorithm reliably produces high-quality parameterizations typically within 3−4 Newton-iterations, even in the presence of extreme aspect-ratios. This makes it particularly attractive for the on-demand application within an automatized industrial MSO-pipeline. To support the demands of modern high-performance computing hardware, only a moderate number of sufficiently large and structured patches is generated which can be mapped one-by-one to the different devices (CPUs/GPUs) with only little communication overhead. Topology changes are avoided in time-dependent and shape-optimization settings. Finally, the spline-based geometry description can be transformed into a classical mesh by performing a large number of function evaluations in the Mapping Operator. Its continuous nature allows for feature-based (structured and unstructured) refinement and arbitrary element densities without increasing the complexity of the meshing process.
-
Elliptic grid generation techniques in the framework of isogeometric analysis applications
'Elsevier BV', 2018Co-Authors: Hinz J.p., Möller M., Vuik CornelisAbstract:The generation of an analysis-suitable computational grid from a description of no more than its boundaries is a common problem in numerical analysis. Most classical meshing techniques for finite-volume, finite-difference or finite-element applications such as the Advancing Front Method (Schöberl, 1997), Delaunay Triangulation (Triangle, 1996) and elliptic or hyperbolic meshing schemes (Thompson et al., 1998) operate with linear or multi-linear but straight-sided elements for the generation of structured and unstructured meshes, respectively, whereas the generation of high-quality curved meshes is still considered a major challenge. A recent development is the introduction of Isogeometric Analysis (IgA) (Hughes et al., 2005), which can be considered as a natural high-order generalisation of the finite-element method. A description of the geometry Ω¯ is accomplished via a Mapping Operator x:Ωˆ→Ω that maps the unit hypercube in Rn onto an approximation Ω of Ω¯ utilizing a linear combination of higher-order spline functions. The numerical simulation is then carried out in the computational domain Ωˆ via a ‘pull back’ using the Mapping Operator x. The advantage is that the flexibility of higher-order spline-functions usually allows for an accurate description of Ω¯ with much fewer elements which can significantly reduce the computational effort required for this step compared to traditional low-order methods. Furthermore, an analytical description of the geometry can be turned back into a traditional (structured or unstructured) grid by performing a large number of function evaluations in x. This can, for instance, be utilized for local refinement without the need for remeshing. A potential drawback of curved instead of linear elements is that the meshing techniques required for the creation of folding-free Mappings tend to be more sophisticated and that it is a less trivial task to verify that the resulting Mapping is indeed bijective. For the purpose of creating folding-free Mappings utilizing spline-functions, we will present an algorithm adopting the principles of Elliptic Grid Generation (EGG) whose basic principles have been adapted to meet the needs of IgA. In R2, EGG has particularly appealing properties since bijectivity of the resulting Mapping is guaranteed as long as the numerical accuracy to the computational approach of the Mapping is sufficient. We will present an algorithm that is capable of generating folding-free Mappings from a large number of geometry contours, including complicated geometries from industrial applications with extreme aspect ratios. This is accomplished by combining EGG with automatized reparameterization techniques and a sophisticated numerical approach for solving resulting governing (nonlinear) equations. The algorithm is equipped with the means to verify the bijectivity of the resulting Mapping and with automatized defect-correction methods in case a violation of bijectivity is detected. Furthermore, we will present possible strategies for the generation of folding-free Mappings for certain types of volumetric geometries by combining EGG with transfinite-interpolation as well as a number of other applications such as time-dependent settings. All applications are provided with example geometries.Green Open Access added to TU Delft Institutional Repository ‘You share, we take care!’ – Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.Numerical Analysi
Hinz J.p. - One of the best experts on this subject based on the ideXlab platform.
-
An IGA Framework for PDE-Based Planar Parameterization on Convex Multipatch Domains
'Springer Science and Business Media LLC', 2021Co-Authors: Hinz J.p., Vuik Cornelis, Möller M., Vuik C., Van Brummelen H., Verhoosel C., Simeon B.Abstract:The first step towards applying isogeometric analysis techniques to solve PDE problems on a given domain consists in generating an analysis-suitable Mapping Operator between parametric and physical domains with one or several patches from no more than a description of the boundary contours of the physical domain. A subclass of the multitude of the available parameterization algorithms are those based on the principles of Elliptic Grid Generation (EGG) which, in their most basic form, attempt to approximate a Mapping Operator whose inverse is composed of harmonic functions. The main challenge lies in finding a formulation of the problem that is suitable for a computational approach and a common strategy is to approximate the Mapping Operator by means of solving a PDE-problem. PDE-based EGG is well-established in classical meshing and first generalization attempts to spline-based descriptions (as is mandatory in IgA) have been made. Unfortunately, all of the practically viable PDE-based approaches impose certain requirements on the employed spline-basis, in particular global C≥1-continuity.This paper discusses an EGG-algorithm for the generation of planar parameterizations with locally reduced smoothness (i.e., with support for locally only C0-continuous bases). A major use case of the proposed algorithm is that of multipatch parameterizations, made possible by the support of C0-continuities. This paper proposes a specially-taylored solution algorithm that exploits many characteristics of the PDE-problem and is suitable for large-scale applications. It is discussed for the single-patch case before generalizing its concepts to multipatch settings. This paper is concluded with three numerical experiments and a discussion of the results
-
An IGA Framework for PDE-Based Planar Parameterization on Convex Multipatch Domains
'Springer Science and Business Media LLC', 2021Co-Authors: Hinz J.p., Möller M., Vuik CornelisAbstract:The first step towards applying isogeometric analysis techniques to solve PDE problems on a given domain consists in generating an analysis-suitable Mapping Operator between parametric and physical domains with one or several patches from no more than a description of the boundary contours of the physical domain. A subclass of the multitude of the available parameterization algorithms are those based on the principles of Elliptic Grid Generation (EGG) which, in their most basic form, attempt to approximate a Mapping Operator whose inverse is composed of harmonic functions. The main challenge lies in finding a formulation of the problem that is suitable for a computational approach and a common strategy is to approximate the Mapping Operator by means of solving a PDE-problem. PDE-based EGG is well-established in classical meshing and first generalization attempts to spline-based descriptions (as is mandatory in IgA) have been made. Unfortunately, all of the practically viable PDE-based approaches impose certain requirements on the employed spline-basis, in particular global C≥1-continuity.This paper discusses an EGG-algorithm for the generation of planar parameterizations with locally reduced smoothness (i.e., with support for locally only C0-continuous bases). A major use case of the proposed algorithm is that of multipatch parameterizations, made possible by the support of C0-continuities. This paper proposes a specially-taylored solution algorithm that exploits many characteristics of the PDE-problem and is suitable for large-scale applications. It is discussed for the single-patch case before generalizing its concepts to multipatch settings. This paper is concluded with three numerical experiments and a discussion of the results.Green Open Access added to TU Delft Institutional Repository ‘You share, we take care!’ – Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.Numerical Analysi
-
Spline-based meshing techniques for industrial applications
International Centre for Numerical Methods in Engineering CIMNE, 2020Co-Authors: Hinz J.p., Mölle M., Vuik CornelisAbstract:Isogeometric Analysis (IgA) has become an accepted framework for the mod-elling, simulation and optimization (MSO) of engineering processes. However, the fully automatized generation of analysis-suitable parameterizations of geometries as they arise in practical workflows is still a challenging task, which often requires application-specific parameterization approaches. In this article we present a practical approach [6] based on the principles of Elliptic Grid Generation (EGG) for the efficient on-demand generation of analysis-suitable spline-based parameterizations. Starting from a (point cloud) description of the boundary provided by the existing MSO-pipeline, an inverse nonlinear Poisson-type problem is solved to obtain a folding-free (planar) parameterization of the entire domain. The non-linearity is efficiently treated with a globalized hierarchical Newton approach. Automatized boundary contour reparameterization techniques are employed to improve the parametric properties from a numerical viewpoint, such as orthogonal isolines and equally-sized cells. The use of curved instead of straight-sided elements allows us to arrive at an accurate description of the target domain with fewer elements and thus potentially lower computational effort. Numerical experiments with screw-compressor geometries demonstrate that the proposed algorithm reliably produces high-quality parameterizations typically within 3−4 Newton-iterations, even in the presence of extreme aspect-ratios. This makes it particularly attractive for the on-demand application within an automatized industrial MSO-pipeline. To support the demands of modern high-performance computing hardware, only a moderate number of sufficiently large and structured patches is generated which can be mapped one-by-one to the different devices (CPUs/GPUs) with only little communication overhead. Topology changes are avoided in time-dependent and shape-optimization settings. Finally, the spline-based geometry description can be transformed into a classical mesh by performing a large number of function evaluations in the Mapping Operator. Its continuous nature allows for feature-based (structured and unstructured) refinement and arbitrary element densities without increasing the complexity of the meshing process.Green Open Access added to TU Delft Institutional Repository ‘You share, we take care!’ – Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.Numerical Analysi
-
Spline-based meshing techniques for industrial applications
International Centre for Numerical Methods in Engineering CIMNE, 2020Co-Authors: Hinz J.p., Vuik Cornelis, Möller M., Owen Roger, De Borst Rene, Reese Jason, Pearce ChrisAbstract:Isogeometric Analysis (IgA) has become an accepted framework for the mod-elling, simulation and optimization (MSO) of engineering processes. However, the fully automatized generation of analysis-suitable parameterizations of geometries as they arise in practical workflows is still a challenging task, which often requires application-specific parameterization approaches. In this article we present a practical approach [6] based on the principles of Elliptic Grid Generation (EGG) for the efficient on-demand generation of analysis-suitable spline-based parameterizations. Starting from a (point cloud) description of the boundary provided by the existing MSO-pipeline, an inverse nonlinear Poisson-type problem is solved to obtain a folding-free (planar) parameterization of the entire domain. The non-linearity is efficiently treated with a globalized hierarchical Newton approach. Automatized boundary contour reparameterization techniques are employed to improve the parametric properties from a numerical viewpoint, such as orthogonal isolines and equally-sized cells. The use of curved instead of straight-sided elements allows us to arrive at an accurate description of the target domain with fewer elements and thus potentially lower computational effort. Numerical experiments with screw-compressor geometries demonstrate that the proposed algorithm reliably produces high-quality parameterizations typically within 3−4 Newton-iterations, even in the presence of extreme aspect-ratios. This makes it particularly attractive for the on-demand application within an automatized industrial MSO-pipeline. To support the demands of modern high-performance computing hardware, only a moderate number of sufficiently large and structured patches is generated which can be mapped one-by-one to the different devices (CPUs/GPUs) with only little communication overhead. Topology changes are avoided in time-dependent and shape-optimization settings. Finally, the spline-based geometry description can be transformed into a classical mesh by performing a large number of function evaluations in the Mapping Operator. Its continuous nature allows for feature-based (structured and unstructured) refinement and arbitrary element densities without increasing the complexity of the meshing process.
-
Elliptic grid generation techniques in the framework of isogeometric analysis applications
2018Co-Authors: Hinz J.p., Möller M., Vuik C.Abstract:The generation of an analysis-suitable computational grid from a description of no more than its boundaries is a common problem in numerical analysis. Most classical meshing techniques for finite-volume, finite-difference or finite-element applications such as the Advancing Front Method (Schöberl, 1997), Delaunay Triangulation (Triangle, 1996) and elliptic or hyperbolic meshing schemes (Thompson et al., 1998) operate with linear or multi-linear but straight-sided elements for the generation of structured and unstructured meshes, respectively, whereas the generation of high-quality curved meshes is still considered a major challenge. A recent development is the introduction of Isogeometric Analysis (IgA) (Hughes et al., 2005), which can be considered as a natural high-order generalisation of the finite-element method. A description of the geometry Ω¯ is accomplished via a Mapping Operator x:Ωˆ→Ω that maps the unit hypercube in Rn onto an approximation Ω of Ω¯ utilizing a linear combination of higher-order spline functions. The numerical simulation is then carried out in the computational domain Ωˆ via a ‘pull back’ using the Mapping Operator x. The advantage is that the flexibility of higher-order spline-functions usually allows for an accurate description of Ω¯ with much fewer elements which can significantly reduce the computational effort required for this step compared to traditional low-order methods. Furthermore, an analytical description of the geometry can be turned back into a traditional (structured or unstructured) grid by performing a large number of function evaluations in x. This can, for instance, be utilized for local refinement without the need for remeshing. A potential drawback of curved instead of linear elements is that the meshing techniques required for the creation of folding-free Mappings tend to be more sophisticated and that it is a less trivial task to verify that the resulting Mapping is indeed bijective. For the purpose of creating folding-free Mappings utilizing spline-functions, we will present an algorithm adopting the principles of Elliptic Grid Generation (EGG) whose basic principles have been adapted to meet the needs of IgA. In R2, EGG has particularly appealing properties since bijectivity of the resulting Mapping is guaranteed as long as the numerical accuracy to the computational approach of the Mapping is sufficient. We will present an algorithm that is capable of generating folding-free Mappings from a large number of geometry contours, including complicated geometries from industrial applications with extreme aspect ratios. This is accomplished by combining EGG with automatized reparameterization techniques and a sophisticated numerical approach for solving resulting governing (nonlinear) equations. The algorithm is equipped with the means to verify the bijectivity of the resulting Mapping and with automatized defect-correction methods in case a violation of bijectivity is detected. Furthermore, we will present possible strategies for the generation of folding-free Mappings for certain types of volumetric geometries by combining EGG with transfinite-interpolation as well as a number of other applications such as time-dependent settings. All applications are provided with example geometries.
Hitoshi Kiya - One of the best experts on this subject based on the ideXlab platform.
-
itm net deep inverse tone Mapping using novel loss function based on tone Mapping Operator
European Signal Processing Conference, 2019Co-Authors: Yuma Kinoshita, Hitoshi KiyaAbstract:A novel inverse tone Mapping network, called “iTMNet”, is proposed in this paper. For training iTM-Net, we also propose a novel loss function considering the pixel distribution of HDR images. In inverse tone Mapping with CNNs, we first point out that training CNNs with a standard loss function causes a problem, due to the distribution of HDR images. To overcome the problem, the novel loss function non-linearly tone-maps target HDR images into LDR ones, on the basis of a tone Mapping Operator, and then the distance between the tone-mapped image and a predicted one is calculated. The proposed loss function enables us not only to normalize HDR images but also to distribute pixel values of HDR images, like LDR ones. Experimental results show that HDR images predicted by the proposed iTM-Net have higher-quality than HDR ones predicted by conventional inverse tone Mapping methods including state-of the-arts, in terms of both HDR-VDP-2.2 and PU encoding + MSSSIM. In addition, compared with loss functions not considering the HDR pixel distribution, the proposed loss function is shown to improve the performance of CNNs.
-
itm net deep inverse tone Mapping using novel loss function considering tone Mapping Operator
IEEE Access, 2019Co-Authors: Yuma Kinoshita, Hitoshi KiyaAbstract:In this paper, we propose a novel inverse tone Mapping network, called “iTM-Net.” For training iTM-Net, we also propose a novel loss function that considers the non-linear relation between low dynamic range (LDR) and high dynamic range (HDR) images. For inverse tone Mapping with convolutional neural networks (CNNs), we first point out that training CNNs with a standard loss function causes a problem due to the non-linear relation between the LDR and HDR images. To overcome the problem, the novel loss function non-linearly tone-maps target HDR images into LDR ones on the basis of a tone Mapping Operator, and the distance between the tone-mapped images and predicted ones are then calculated. The proposed loss function enables us not only to normalize the HDR images but also to reduce the non-linear relation between LDR and HDR ones. The experimental results show that the HDR images predicted by the proposed iTM-Net have higher-quality than the HDR ones predicted by conventional inverse tone Mapping methods, including the state of the art, in terms of both HDR-VDP-2.2 and PU encoding + MS-SSIM. In addition, compared with loss functions that do not consider the non-linear relation, the proposed loss function is shown to improve the performance of CNNs.
-
reinhard s global Operator based inverse tone Mapping with one parameter
International Workshop on Signal Design and its Applications in Communications, 2017Co-Authors: Yuma Kinoshita, Sayaka Shiota, Hitoshi KiyaAbstract:This paper proposes a new inverse tone Mapping Operator (TMO) with only one parameter. The proposed inverse TMO is based on Reinhard's global Operator which is a well-known TMO. Inverse TM operations have two applications: generating an HDR image from an existing LDR one, and reconstructing an original HDR image from the mapped LDR image. In this paper, we focus on the latter application and the proposed method can produce an HDR image that has high image quality. In the latter application, two parameters used in Reinhard's TMO, i.e. the key value α regarding brightness of a mapped LDR one and the geometric mean L w of an original HDR one, are generally required for carrying out the Reinhard based inverse_TMO. In this paper, we show that it is possible to estimate L w from α under some conditions, so that a new inverse TMO with one parameter is proposed. Experimental results show that the proposed method outperforms conventional ones for the latter application, in terms of high image quality and low computing cost.
-
an inverse tone Mapping Operator based on reinhard s global Operator
International Symposium on Intelligent Signal Processing and Communication Systems, 2016Co-Authors: Yuma Kinoshita, Sayaka Siota, Hitoshi KiyaAbstract:A number of inverse tone Mapping Operators (TMOs) for dynamic range expansion have been proposed due to the need to visualize low dynamic range (LDR) images on high dynamic range (HDR) devices. This paper proposes a novel inverse TMO, which enables to generate HDR images from LDR ones, not only without using any specific parameters but also at low computing costs. Furthermore, the inverse TMO has a new characteristic when an LDR image is mapped from an HDR one by Reinhard's global Operator. In the case, the HDR image reconstructed by the proposed method without parameters can be remapped into the same image as that remapped from an HDR one reconstructed with parameters. Experimental results show that the proposed inverse tone Mapping (TM) operation can be carried out, while keeping better structural similarity and lower computing cost than conventional methods.
Giuseppe Valenzise - One of the best experts on this subject based on the ideXlab platform.
-
Deep Tone Mapping Operator for High Dynamic Range Images
IEEE Transactions on Image Processing, 2019Co-Authors: Aakanksha Rana, Praveer Singh, Giuseppe Valenzise, Frédéric Dufaux, Nikos Komodakis, Aljosa SmolicAbstract:A computationally fast tone Mapping Operator (TMO) that can quickly adapt to a wide spectrum of high dynamic range (HDR) content is quintessential for visualization on varied low dynamic range (LDR) output devices such as movie screens or standard displays. Existing TMOs can successfully tone-map only a limited number of HDR content and require an extensive parameter tuning to yield the best subjective-quality tone-mapped output. In this paper, we address this problem by proposing a fast, parameter-free and scene-adaptable deep tone Mapping Operator (DeepTMO) that yields a high-resolution and high-subjective quality tone mapped output. Based on conditional generative adversarial network (cGAN), DeepTMO not only learns to adapt to vast scenic-content (e.g., outdoor, indoor, human, structures, etc.) but also tackles the HDR related scene-specific challenges such as contrast and brightness, while preserving the fine-grained details. We explore 4 possible combinations of Generator-Discriminator architectural designs to specifically address some prominent issues in HDR related deep-learning frameworks like blurring, tiling patterns and saturation artifacts. By exploring different influences of scales, loss-functions and normalization layers under a cGAN setting, we conclude with adopting a multi-scale model for our task. To further leverage on the large-scale availability of unlabeled HDR data, we train our network by generating targets using an objective HDR quality metric, namely Tone Mapping Image Quality Index (TMQI). We demonstrate results both quantitatively and qualitatively, and showcase that our DeepTMO generates high-resolution, high-quality output images over a large spectrum of real-world scenes. Finally, we evaluate the perceived quality of our results by conducting a pair-wise subjective study which confirms the versatility of our method.
-
learning based tone Mapping Operator for efficient image matching
IEEE Transactions on Multimedia, 2019Co-Authors: Aakanksha Rana, Giuseppe Valenzise, Frédéric DufauxAbstract:In this paper, we propose a new framework to optimally tone map the high dynamic range (HDR) content for image matching under drastic illumination variations. Since tone Mapping Operators (TMO) have traditionally been used for displaying HDR scenes, their design is suboptimal when used for computer vision tasks, such as image matching. We address this suboptimality by proposing a two-step framework, consisting of: first, a luminance-invariant guidance model based on a support vector regressor (SVR) to optimally adapt the tone Mapping function for image matching; and second, an energy maximization model to generate appropriate training samples for learning the SVR. At each step, we collectively address both stages of keypoint detection and descriptor extraction in the feature matching framework. By locally altering the intrinsic characteristics of the tone Mapping function, the learned guidance model facilitates the extraction of local invariant features in the presence of illumination variations. We demonstrate that the proposed TMO significantly outperforms perceptually driven state-of-the-art TMOs on a dataset of HDR scenes characterized by challenging lighting variations, such as day/night transitions.
-
spatio temporal constrained tone Mapping Operator for hdr video compression
Journal of Visual Communication and Image Representation, 2018Co-Authors: Giuseppe Valenzise, Cagri Ozcinar, Paul Lauga, Frédéric DufauxAbstract:Abstract With the growing popularity of high dynamic range (HDR) imaging, efficient compression techniques are demanded, as HDR video entails typically higher raw data rate than traditional video. For this purpose, we introduce a hybrid spatially and temporally constrained content-adaptive tone Mapping Operator (TMO) to convert the input HDR video into a tone mapped video sequence, which is then encoded using the high efficiency video coding (HEVC) standard. The proposed TMO simultaneously exploits intra-frame spatial redundancies and preserves inter-frame temporal coherence of the tone mapped video sequence. Extensive experimental results show that the developed spatio-temporal TMO (ST-TMO) solution yields higher coding performance than existing frame-by-frame TMO’s, and compares favorably with state-of-the-art methods based on a fixed transfer function.
-
Learning-Based Tone Mapping Operator for Image Matching
2017Co-Authors: Aakanksha Rana, Giuseppe Valenzise, Frédéric DufauxAbstract:In this paper, we propose a new framework to optimally tone-map a high dynamic range (HDR) content for image matching under drastic illumination variations. This task is of fundamental importance for many computer vision applications. To design such a framework, we build a luminance invariant guidance model using a Support Vector Regressor (SVR) and learn it to facilitate the extraction of invariant descriptors from scenes subject to wide variety of appearance changes such as day/night transition. To this end, we initially generate appropriate training samples using a simple similarity-maximization mechanism. We then employ the learned model to predict optimal modulation maps that help to locally alter the intrinsic characteristics (such as shape, size) of the tone Mapping function. We evaluate the proposed model performance in terms of matching score and mean average precision rate using state-of-the-art descriptor extraction schemes. We demonstrate that our tone Mapping framework significantly outperforms the existing perceptually-driven state-of-the-art TMOs on the benchmark datasets.
-
HDR Video Coding based on a Temporally Constrained Tone Mapping Operator,
2016Co-Authors: Cagri Ozcinar, Giuseppe Valenzise, Paul Lauga, Frédéric DufauxAbstract:Given its potential for more realistic rendering and enhanced user experience, High Dynamic Range (HDR) imaging is raising a lot of interest both in industry and academia. In this context, efficient representation and coding techniques are needed, as HDR video entails significantly higher raw data rate. In this paper, we present a temporally constrained contentadaptive Tone Mapping Operator (TMO) in order to convert the input HDR video into a reduced bit depth video sequence which is then encoded using High Efficiency Video Coding (HEVC). As the proposed TMO simultaneously takes into account the statistical characteristics of the input frame while better preserving temporal coherence of the tone mapped video sequence, it leads to improved coding efficiency. Experimental results show that the proposed technique compares favorably with existing methods in terms of rate-distortion when using the HDR-VDP-2.2.1 quality metric.