Every year, several hundred thousand people suffer a stroke often leading to long-term motor disa... more Every year, several hundred thousand people suffer a stroke often leading to long-term motor disabilities that impair their quality of life. In this context, hemiplegia including paralysis of hand and fingers plays a key role, leaving stroke survivors unable to perform tasks requiring both hands. In case of lesions at the level of the brain stem or the spinal cord, paralysis can also affect both sides resulting in very severe constraints for performing most activities of daily living. A neural-guided hand exoskeleton can restore motor hand function after a stroke or spinal cord injury. However, controlling such hand exoskeleton raises several challenges related to human-machine interaction. While it should be operated without the user's hands and require as little physical and cognitive strain on them as possible, it should be also as inconspicuous as possible to avoid stigmatization of the users. To tackle these challenges, we conducted a survey among 62 healthy test persons to...
Twelfth International Conference on Machine Vision (ICMV 2019), 2020
In this paper, we present a novel 3D scene reconstruction framework from a single front-mounted s... more In this paper, we present a novel 3D scene reconstruction framework from a single front-mounted stereo camera on a moving vehicle. We propose image triangulations to efficiently render a 3D scene only from 2D textures, while introducing tube meshes as an effective way to render out-of-frustum points. Furthermore, we derive a 3D extended Kalman filter to fuse stereo estimates temporally between frames and showcase a render pipeline, which exploits OpenGL shaders to offload computational costs from the CPU to the GPU. Our approach is able to increase the stereo accuracy compared to competing approaches on the KITTI visual odometry dataset. We also introduce a challenging view prediction evaluation scenario on the SYNTHIA dataset, in which our approach comes out on top in terms of SSIM, 1-NCC error and completeness.
2017 IEEE International Conference on Robotics and Automation (ICRA), 2017
We present LS-ELAS, a line segment extension to the ELAS algorithm, which increases the performan... more We present LS-ELAS, a line segment extension to the ELAS algorithm, which increases the performance and robustness. LS-ELAS is a binocular dense stereo matching algorithm, which computes the disparities in constant time for most of the pixels in the image and in linear time for a small subset of the pixels (support points). Our approach is based on line segments to determine the support points instead of uniformly selecting them over the image range. This way we find very informative support points which preserve the depth discontinuity. The prior of our Bayesian stereo matching method is based on a set of line segments and a set of support points. Both sets are plugged into a constrained Delaunay triangulation to generate a triangulation mesh which is aware of possible depth discontinuities. We further increased the accuracy by using an adaptive method to sample candidate points along edge segments.
This paper deals with the problem of simplifying, transmitting and rendering large textured terra... more This paper deals with the problem of simplifying, transmitting and rendering large textured terrain models. The terrain is rendered from a hierarchy of quadratic tiles of geometry. Each of these tiles has a corresponding texture with a constant number of texels, e.g. 64x64, in each resolution level. Therefore, the ratio between the needed geometric accuracy within a tile and the texel size is a constant, independent of the level of detail, e.g. one texel (1/64 of the width of the tile). This observation allows us to create an efficient new multiresolution model for terrain data where not only the number of vertices is adapted to the level of detail but also the relative accuracy of the coordinates. Although we do not use a pure quadtree data structure due to its known problems with the representation of not axis aligned geometric features, our data structure is very compact and avoids redundant transmission of coordinates.
Proceedings of the 18th annual conference on Computer graphics and interactive techniques - SIGGRAPH '91, 1991
Antialiasing of edges is often performed with the help of subpixel masks that indicate which part... more Antialiasing of edges is often performed with the help of subpixel masks that indicate which parts of the pixel are covered by the object that has to be drawn. For this purpose, subpixel masks have to be generated during the scan conversion of an image. This paper introduces a new algorithm for creating subpixel masks that avoids some problems of traditional algorithms, like aliasing of high frequencies or blinking of small moving objects. The new algorithm can be implemented by lookup tables that make use of the inherent symmetry of the algorithm. The results are compared with conventional supersampling 1. A hardware implementation is described.
Every year, several hundred thousand people suffer a stroke often leading to long-term motor disa... more Every year, several hundred thousand people suffer a stroke often leading to long-term motor disabilities that impair their quality of life. In this context, hemiplegia including paralysis of hand and fingers plays a key role, leaving stroke survivors unable to perform tasks requiring both hands. In case of lesions at the level of the brain stem or the spinal cord, paralysis can also affect both sides resulting in very severe constraints for performing most activities of daily living. A neural-guided hand exoskeleton can restore motor hand function after a stroke or spinal cord injury. However, controlling such hand exoskeleton raises several challenges related to human-machine interaction. While it should be operated without the user's hands and require as little physical and cognitive strain on them as possible, it should be also as inconspicuous as possible to avoid stigmatization of the users. To tackle these challenges, we conducted a survey among 62 healthy test persons to...
Twelfth International Conference on Machine Vision (ICMV 2019), 2020
In this paper, we present a novel 3D scene reconstruction framework from a single front-mounted s... more In this paper, we present a novel 3D scene reconstruction framework from a single front-mounted stereo camera on a moving vehicle. We propose image triangulations to efficiently render a 3D scene only from 2D textures, while introducing tube meshes as an effective way to render out-of-frustum points. Furthermore, we derive a 3D extended Kalman filter to fuse stereo estimates temporally between frames and showcase a render pipeline, which exploits OpenGL shaders to offload computational costs from the CPU to the GPU. Our approach is able to increase the stereo accuracy compared to competing approaches on the KITTI visual odometry dataset. We also introduce a challenging view prediction evaluation scenario on the SYNTHIA dataset, in which our approach comes out on top in terms of SSIM, 1-NCC error and completeness.
2017 IEEE International Conference on Robotics and Automation (ICRA), 2017
We present LS-ELAS, a line segment extension to the ELAS algorithm, which increases the performan... more We present LS-ELAS, a line segment extension to the ELAS algorithm, which increases the performance and robustness. LS-ELAS is a binocular dense stereo matching algorithm, which computes the disparities in constant time for most of the pixels in the image and in linear time for a small subset of the pixels (support points). Our approach is based on line segments to determine the support points instead of uniformly selecting them over the image range. This way we find very informative support points which preserve the depth discontinuity. The prior of our Bayesian stereo matching method is based on a set of line segments and a set of support points. Both sets are plugged into a constrained Delaunay triangulation to generate a triangulation mesh which is aware of possible depth discontinuities. We further increased the accuracy by using an adaptive method to sample candidate points along edge segments.
This paper deals with the problem of simplifying, transmitting and rendering large textured terra... more This paper deals with the problem of simplifying, transmitting and rendering large textured terrain models. The terrain is rendered from a hierarchy of quadratic tiles of geometry. Each of these tiles has a corresponding texture with a constant number of texels, e.g. 64x64, in each resolution level. Therefore, the ratio between the needed geometric accuracy within a tile and the texel size is a constant, independent of the level of detail, e.g. one texel (1/64 of the width of the tile). This observation allows us to create an efficient new multiresolution model for terrain data where not only the number of vertices is adapted to the level of detail but also the relative accuracy of the coordinates. Although we do not use a pure quadtree data structure due to its known problems with the representation of not axis aligned geometric features, our data structure is very compact and avoids redundant transmission of coordinates.
Proceedings of the 18th annual conference on Computer graphics and interactive techniques - SIGGRAPH '91, 1991
Antialiasing of edges is often performed with the help of subpixel masks that indicate which part... more Antialiasing of edges is often performed with the help of subpixel masks that indicate which parts of the pixel are covered by the object that has to be drawn. For this purpose, subpixel masks have to be generated during the scan conversion of an image. This paper introduces a new algorithm for creating subpixel masks that avoids some problems of traditional algorithms, like aliasing of high frequencies or blinking of small moving objects. The new algorithm can be implemented by lookup tables that make use of the inherent symmetry of the algorithm. The results are compared with conventional supersampling 1. A hardware implementation is described.
Uploads
Papers by Andreas Schilling