Time filter

Source Type

Le Moan S.,TU Darmstadt | Urban P.,Fraunhofer Institute for Computer Graphics Research
IEEE Transactions on Image Processing | Year: 2014

We propose a new strategy to evaluate the quality of multi and hyperspectral images, from the perspective of human perception. We define the spectral image difference as the overall perceived difference between two spectral images under a set of specified viewing conditions (illuminants). First, we analyze the stability of seven image-difference features across illuminants, by means of an information-theoretic strategy. We demonstrate, in particular, that in the case of common spectral distortions (spectral gamut mapping, spectral compression, spectral reconstruction), chromatic features vary much more than achromatic ones despite considering chromatic adaptation. Then, we propose two computationally efficient spectral image difference metrics and compare them to the results of a subjective visual experiment. A significant improvement is shown over existing metrics such as the widely used root-mean square error. © 2014 IEEE.

Preiss J.,TU Darmstadt | Fernandes F.,TU Darmstadt | Urban P.,Fraunhofer Institute for Computer Graphics Research
IEEE Transactions on Image Processing | Year: 2014

While image-difference metrics show good prediction performance on visual data, they often yield artifact-contaminated results if used as objective functions for optimizing complex image-processing tasks. We investigate in this regard the recently proposed color-image-difference (CID) metric particularly developed for predicting gamut-mapping distortions. We present an algorithm for optimizing gamut mapping employing the CID metric as the objective function. Resulting images contain various visual artifacts, which are addressed by multiple modifications yielding the improved color-image-difference (iCID) metric. The iCID-based optimizations are free from artifacts and retain contrast, structure, and color of the original image to a great extent. Furthermore, the prediction performance on visual data is improved by the modifications. © 2013 IEEE.

Steger S.,Fraunhofer Institute for Computer Graphics Research
Medical image computing and computer-assisted intervention : MICCAI ... International Conference on Medical Image Computing and Computer-Assisted Intervention | Year: 2012

This paper presents a novel skeleton based method for the registration of head&neck datasets. Unlike existing approaches it is fully automated, spatial relation of the bones is considered during their registration and only one of the images must be a CT scan. An articulated atlas is used to jointly obtain a segmentation of the skull, the mandible and the vertebrae C1-Th2 from the CT image. These bones are then successively rigidly registered with the moving image, beginning at the skull, resulting in a rigid transformation for each of the bones. Linear combinations of those transformations describe the deformation in the soft tissue. The weights for the transformations are given by the solution of the Laplace equation. Optionally, the skin surface can be incorporated. The approach is evaluated on 20 CT/MRI pairs of head&neck datasets acquired in clinical routine. Visual inspection shows that the segmentation of the bones was successful in all cases and their successive alignment was successful in 19 cases. Based on manual segmentations of lymph nodes in both modalities, the registration accuracy in the soft tissue was assessed. The mean target registration error of the lymph node centroids was 5.33 +/- 2.44 mm when the registration was solely based on the deformation of the skeleton and 5.00 +/- 2.38 mm when the skin surface was additionally considered. The method's capture range is sufficient to cope with strongly deformed images and it can be modified to support other parts of the body. The overall registration process typically takes less than 2 minutes.

Seibert H.,Fraunhofer Institute for Computer Graphics Research
Proceedings of the 2012 8th International Conference on Intelligent Information Hiding and Multimedia Signal Processing, IIH-MSP 2012 | Year: 2012

The segmentation of the face in 3D reconstructions is a crucial processing step within 3D face recognition systems. At this early processing stage discarding other surface portions such as collars, hats or hairstyle shall reduce the amount of data. In contrast to other approaches, the proposed algorithm uses only the face geometry and is therefore robust with respect to lighting conditions or texture quality. Assuming the skin region of a face is locally flat and closed, a binary mask image is created. Morphology and a simple heuristic are applied on connected components to select and join appropriate components. The implementation is straight forward, yielding just a few parameters and copes the problem without training procedure. A proof of concept is given and results are shown for several cases, limitations of the approach are discussed. © 2012 IEEE.

Steger S.,Fraunhofer Institute for Computer Graphics Research | Sakas G.,Fraunhofer Institute for Computer Graphics Research
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2012

Automatic segmentation methods for tumors are typically only suitable for a specific type of tumor in a specific imaging modality and sometimes lack in accuracy whereas manual tumor segmentation achieves the desired results but is very time consuming. Interactive segmentation however speeds up the process while still being able to maintain the accuracy of manual segmentation. This paper presents a novel method for fast interactive segmentation of tumors (called FIST) from medical images, which is suitable for all somewhat spherical tumors in any 3d medical imaging modality. The user clicks in the center of the tumor and a belief propagation based iterative adaption process is initiated, thereby considering image gradients as well as local smoothness priors of the surface. During that process, instant visual feedback is given, enabling to intervene in the adaption process by sketching parts of the contour in any cross section. The approach has successfully been applied to the segmentation of liver tumors in CT datasets. Satisfactory results could be achieved in 15.20875 seconds on the average. Further trials on oropharynx tumors, liver tumors and the prostate from MR images as well as lymph nodes and the bladder from CT volumes demonstrate the generality of the presented approach. © 2012 Springer-Verlag.

Tazari M.-R.,Fraunhofer Institute for Computer Graphics Research
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2010

Challenges of handling user interaction in Ambient Intelligence environments are manifold. The systems installed in these environments are highly distributed with dynamic configurations in terms of integrated devices and installed applications. Context-awareness, personalization, and multimodality are critical for supporting more natural interaction and optimizing the interaction in an adaptive way. Research activities have dealt with different specific problems in the field and now it is high time for moving towards an open framework with a more comprehensive solution. This paper introduces results of such work with a high level of freedom in developing and deploying applications without needing to care about the available I/O infrastructure. The other way around, the latter can be changed without worrying about the application side. The independence of applications from the available I/O infrastructure helps to share mechanisms and manage such a complex scene more adequately. The key idea behind this framework is the natural distribution of tasks according to the real scene using a middleware solution supporting seamless connectivity and goal-based interoperability. © 2010 Springer-Verlag Berlin Heidelberg.

Reitz T.,Fraunhofer Institute for Computer Graphics Research
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2010

Geospatial data offered by distributed services are often modeled with different conceptual schemas although they cover the same thematic area. To ensure interoperability of geospatial data, the existing heterogeneous conceptual schemas can be mapped to a common conceptual schema. However, the underlying formalized schema mappings are difficult to create, difficult to re-use and often contain mismatches of abstraction level, of scope difference, domain semantics and value semantics of the mapped entities. We have developed a novel approach to document and communicate such mismatches in the form of a Mismatch Description Language (MDL). This MDL can be transformed into various textual and cartographic representations to support users in communicating and understanding mismatches, and to assess the reusability of a mapping. © 2010 Springer-Verlag.

Kahn S.,Fraunhofer Institute for Computer Graphics Research
Virtual Reality | Year: 2013

Whereas 3D surface models are often used for augmented reality (e. g., for occlusion handling or model-based camera tracking), the creation and the use of such dense 3D models in augmented reality applications usually are two separated processes. The 3D surface models are often created in offline preparation steps, which makes it difficult to detect changes and to adapt the 3D model to these changes. This work presents a 3D change detection and model adjustment framework that combines AR techniques with real-time depth imaging to close the loop between dense 3D modeling and augmented reality. The proposed method detects the differences between a scene and a 3D model of the scene in real time. Then, the detected geometric differences are used to update the 3D model, thus bringing AR and 3D modeling closer together. The accuracy of the geometric difference detection depends on the depth measurement accuracy as well as on the accuracy of the intrinsic and extrinsic parameters. To evaluate the influence of these parameters, several experiments were conducted with simulated ground truth data. Furthermore, the evaluation shows the applicability of AR and depth image-based 3D modeling for model-based camera tracking. © 2011 Springer-Verlag London Limited.

Olbrich M.,Fraunhofer Institute for Computer Graphics Research
Proceedings, Web3D 2012 - 17th International Conference on 3D Web Technology | Year: 2012

X3D supports a variety of media types to be used in 3D scenes, like images, videos or other X3D models. A scene can dynamically load and replace this media during runtime, but since there is no way to communicate directly with outside sources like a server, all data sources need to be known in advance. This problem is usually solved by using interfaces like SAI, which allow external applications to modify the current scene. But this solution makes it necessary to set up all the communication via SAI and have the external application communicate with a server. In this paper, we will show how XMLHttpRequest, an object common in web browsers, can be used to handle the communication from within the X3D Browser. We will show how well this approach fits into the X3D environment and how easy this can be implemented in an X3D Browser. Afterwards, some examples will show the benefits in real applications and how easy this solution is to use. © 2012 ACM.

Aehnelt M.,Fraunhofer Institute for Computer Graphics Research | Bader S.,University of Rostock
ICAART 2015 - 7th International Conference on Agents and Artificial Intelligence, Proceedings | Year: 2015

Information assistance helps in many application domains to structure, guide and control human work processes. However, it lacks a formalisation and automated processing of background knowledge which vice versa is required to provide ad-hoc assistance. In this paper, we describe our conceptual and technical work to include contextual background knowledge in raising awareness, guiding, and monitoring the assembly worker. We present cognitive architectures as missing link between highly sophisticated manufacturing data systems and implicitly available contextual knowledge on work procedures and concepts of the work domain. Our work is illustrated with examples in SWI-Prolog and the Soar cognitive architecture.

Loading Fraunhofer Institute for Computer Graphics Research collaborators
Loading Fraunhofer Institute for Computer Graphics Research collaborators