Philippos Mordohai | Stevens Institute of Technology (original) (raw)

Papers by Philippos Mordohai

Research paper thumbnail of Towards Urban 3D Reconstruction from Video

Third International Symposium on 3D Data Processing, Visualization, and Transmission (3DPVT'06), 2006

Bookmarks Related papers MentionsView impact

Research paper thumbnail of Learning to Detect Ground Control Points for Improving the Accuracy of Stereo Matching

2014 IEEE Conference on Computer Vision and Pattern Recognition, 2014

Bookmarks Related papers MentionsView impact

Research paper thumbnail of 3D Interest Point Detection via Discriminative Learning

Bookmarks Related papers MentionsView impact

Research paper thumbnail of Automatic facial expression recognition using bags of motion words

Bookmarks Related papers MentionsView impact

Research paper thumbnail of Detailed Real-Time Urban 3D Reconstruction from Video

International Journal of Computer Vision, 2008

Bookmarks Related papers MentionsView impact

Research paper thumbnail of Detecting Patterns in Vector Fields

We describe global and local methods for comparing flow fields, and a visualization tool that all... more We describe global and local methods for comparing flow fields, and a visualization tool that allows the user to adjust how flow fields are compared. Our first global method operates on path-lines and measures variations in orientation and curvature between samples on the ...

Bookmarks Related papers MentionsView impact

Research paper thumbnail of Slip Surfaces in Fault Breccia From the Sierra Madre Fault Zone: Geometry and Mechanical Implications

Fault breccia from the Sierra Madre Fault Zone is imaged using a new technique, resolving a highl... more Fault breccia from the Sierra Madre Fault Zone is imaged using a new technique, resolving a highly detailed structure of internal slip surfaces with characteristic distribution of orientations and senses of slip. We excavated a 25x28x35 cm3 sample of brecciated granite from the hanging wall of the Sierra Madre fault near JPL, La Canada, California. The base of the sample was attached to the principal slip surface, dipping 40° to the NW, showing up-dip slip striations. The sample was carefully removed with known orientation while maintaining its integrity. We used a two camera digital imaging system developed in the Department of Electrical Engineering at USC. With the positions of the cameras and sample fixed, we progressively removed the outer layers of the specimen exposing the internal slip surfaces. Each slip plane was marked with a spot; a line was marked if slicknsides were observed, and a stereo photo taken. These 3D images were analyzed to give the normal to each slip plane and the tangent to each slip line. The slip surfaces are few millimeters to several centimeters wide. They branch and coalesce to create a dense micro- and meso-scale network. The polygons bounded by these surfaces are composed of powdered granite that maintains its original fabric. The surfaces are shiny, present clear slicknsides and have silky touch. Larger slip surfaces show macroscopic curvature producing roughness at various wavelengths. This curvature is comprised of the smaller slip planes that were marked for the imaging. We hypothesize that initial fracturing process created the small-scale planes which then coalescenced by rotation and abrasive wear to create the large surfaces. These large slip surfaces are generally sub-parallel to the fault strike and sub-perpendicular to its slip vector. Analysis of 196 planes indicates that the dip direction of 123 planes (63%) is between the east and the south, nearly opposite to the fault. Another 59 planes are dipping in the NE and SW quadrants, 43 of them trend to the ENE and to the SSW. Only 14 planes (7%) trend NW, sub-parallel to the fault. The distribution of inclinations covers the entire range, except for the planes that trend with the fault. Of these, 9 are sub-vertical and 3 are sub-horizontal. The slicknsides were imaged as lines in space but the senses of motion have not yet been determined. They are low-angle 3D lines with no consistent trend indicating mostly horizontal motion. The mechanical origin of this set of surfaces needs a further verification. While they are not consistent with Riedel shear geometry, the characteristics of the surfaces show that they accommodate a small amount of rotation that may reflect simple shear. This requires a relatively homogeneous stress field. As this is not necessarily the case, they might be of dynamic origin, reflecting the heterogeneity of a process-zone stress field ahead of the earthquake rupture. Finally, the dynamic reduction of normal stress during the passage of a rupture can control the formation of this geometry, but requires that the slicknsides (to be verified) will demonstrate a tensile regime.

Bookmarks Related papers MentionsView impact

Research paper thumbnail of Binocular Stereo

Bookmarks Related papers MentionsView impact

Research paper thumbnail of A perceptual organization approach for figure completion, binocular and multiple-view stereo and machine learning using tensor voting

Bookmarks Related papers MentionsView impact

Research paper thumbnail of Real-time video-based reconstruction of urban environments

Proceedings of 3DARCH: 3D Virtual Reconstruction and Visualization of Complex Architectures, Jul 1, 2007

1University of North Carolina, Chapel Hill, NC, USA {mordohai, jmf, bclipp, gallup, pmerrell, sal... more 1University of North Carolina, Chapel Hill, NC, USA {mordohai, jmf, bclipp, gallup, pmerrell, salmi, taltonb, herman, welch, marc}@ cs. unc. edu 2 University of Kentucky, Lexington, KY, USA {amir, engels, qingxiong. yang, stewe}@ vis. uky. edu,{lwangd, ryang, dnister}@ cs. uky. edu KEY WORDS: computer vision, 3D model, stereo reconstruction, large-scale reconstruction, real time processing, geo-registration

Bookmarks Related papers MentionsView impact

Research paper thumbnail of Binocular and Multiple View Stereo using Tensor Voting

Bookmarks Related papers MentionsView impact

Research paper thumbnail of Inference of Segmented Overlapping Surfaces From Binocular and Multiple-View Stereo

We present an integrated approach,to the derivation of scene descriptions from two or more stereo... more We present an integrated approach,to the derivation of scene descriptions from two or more stereo images, where evaluation of feature correspondences and grouping into perceptual structures are addressed,within the same,framework.,Special attention is given to the development of a methodology with general applicability. In order to handle the issues of noise, lack of image features, surface discontinuities and half occluded

Bookmarks Related papers MentionsView impact

Research paper thumbnail of Dimensionality Estimation and Manifold Learning using Tensor Voting

We address instance-based learning from a perceptual organization standpoint us- ing tensor votin... more We address instance-based learning from a perceptual organization standpoint us- ing tensor voting. The goal of instance-based learning is to learn the underlying relationships between observations, which are points in an N-D continuous space, under the assumption that they lie in a limited part of the N-D space, typically a manifold, the dimensionality of which is an indication of the

Bookmarks Related papers MentionsView impact

Research paper thumbnail of Unsupervised Dimensionality Estimation and Manifold Learning in high-dimensional Spaces by Tensor Voting

International Joint Conference on Artificial Intelligence, 2005

We address dimensionality estimation and nonlin- ear manifold inference starting from point input... more We address dimensionality estimation and nonlin- ear manifold inference starting from point inputs in high dimensional spaces using tensor voting. The proposed method operates locally in neigh- borhoods and does not involve any global com- putations. It is based on information propagation among neighboring points implemented as a voting process. Unlike other local approaches for man- ifold learning, the quantity

Bookmarks Related papers MentionsView impact

Research paper thumbnail of Object Detection from Large-Scale 3D Datasets Using Bottom-Up and Top-Down Descriptors

Lecture Notes in Computer Science, 2008

ABSTRACT We propose an approach for detecting objects in large-scale range datasets that combines... more ABSTRACT We propose an approach for detecting objects in large-scale range datasets that combines bottom-up and top-down processes. In the bottom-up stage, fast-to-compute local descriptors are used to detect potential target objects. The object hypotheses are verified after alignment in a top-down stage using global descriptors that capture larger scale structure information. We have found that the combination of spin images and Extended Gaussian Images, as local and global descriptors respectively, provides a good trade-off between efficiency and accuracy. We present results on real outdoors scenes containing millions of scanned points and hundreds of targets. Our results compare favorably to the state of the art by being applicable to much larger scenes captured under less controlled conditions, by being able to detect object classes and not specific instances, and by being able to align the query with the best matching model accurately, thus obtaining precise segmentation.

Bookmarks Related papers MentionsView impact

Research paper thumbnail of A hybrid control approach to the Next-Best-View problem using stereo vision

2013 IEEE International Conference on Robotics and Automation, 2013

Bookmarks Related papers MentionsView impact

Research paper thumbnail of Tensor Voting: A Perceptual Organization Approach to Computer Vision and Machine Learning

Synthesis Lectures on Image, Video, and Multimedia Processing, 2006

Bookmarks Related papers MentionsView impact

Research paper thumbnail of Multi-View Stereo via Graph Cuts on the Dual of an Adaptive Tetrahedral Mesh

2007 IEEE 11th International Conference on Computer Vision, 2007

Bookmarks Related papers MentionsView impact

Research paper thumbnail of Correspondence estimation for non-rigid point clouds with automatic part discovery

The Visual Computer, 2015

Bookmarks Related papers MentionsView impact

Research paper thumbnail of Classification of Vehicle Parts in Unstructured 3D Point Clouds

2014 2nd International Conference on 3D Vision, 2014

Bookmarks Related papers MentionsView impact

Research paper thumbnail of Towards Urban 3D Reconstruction from Video

Third International Symposium on 3D Data Processing, Visualization, and Transmission (3DPVT'06), 2006

Bookmarks Related papers MentionsView impact

Research paper thumbnail of Learning to Detect Ground Control Points for Improving the Accuracy of Stereo Matching

2014 IEEE Conference on Computer Vision and Pattern Recognition, 2014

Bookmarks Related papers MentionsView impact

Research paper thumbnail of 3D Interest Point Detection via Discriminative Learning

Bookmarks Related papers MentionsView impact

Research paper thumbnail of Automatic facial expression recognition using bags of motion words

Bookmarks Related papers MentionsView impact

Research paper thumbnail of Detailed Real-Time Urban 3D Reconstruction from Video

International Journal of Computer Vision, 2008

Bookmarks Related papers MentionsView impact

Research paper thumbnail of Detecting Patterns in Vector Fields

We describe global and local methods for comparing flow fields, and a visualization tool that all... more We describe global and local methods for comparing flow fields, and a visualization tool that allows the user to adjust how flow fields are compared. Our first global method operates on path-lines and measures variations in orientation and curvature between samples on the ...

Bookmarks Related papers MentionsView impact

Research paper thumbnail of Slip Surfaces in Fault Breccia From the Sierra Madre Fault Zone: Geometry and Mechanical Implications

Fault breccia from the Sierra Madre Fault Zone is imaged using a new technique, resolving a highl... more Fault breccia from the Sierra Madre Fault Zone is imaged using a new technique, resolving a highly detailed structure of internal slip surfaces with characteristic distribution of orientations and senses of slip. We excavated a 25x28x35 cm3 sample of brecciated granite from the hanging wall of the Sierra Madre fault near JPL, La Canada, California. The base of the sample was attached to the principal slip surface, dipping 40° to the NW, showing up-dip slip striations. The sample was carefully removed with known orientation while maintaining its integrity. We used a two camera digital imaging system developed in the Department of Electrical Engineering at USC. With the positions of the cameras and sample fixed, we progressively removed the outer layers of the specimen exposing the internal slip surfaces. Each slip plane was marked with a spot; a line was marked if slicknsides were observed, and a stereo photo taken. These 3D images were analyzed to give the normal to each slip plane and the tangent to each slip line. The slip surfaces are few millimeters to several centimeters wide. They branch and coalesce to create a dense micro- and meso-scale network. The polygons bounded by these surfaces are composed of powdered granite that maintains its original fabric. The surfaces are shiny, present clear slicknsides and have silky touch. Larger slip surfaces show macroscopic curvature producing roughness at various wavelengths. This curvature is comprised of the smaller slip planes that were marked for the imaging. We hypothesize that initial fracturing process created the small-scale planes which then coalescenced by rotation and abrasive wear to create the large surfaces. These large slip surfaces are generally sub-parallel to the fault strike and sub-perpendicular to its slip vector. Analysis of 196 planes indicates that the dip direction of 123 planes (63%) is between the east and the south, nearly opposite to the fault. Another 59 planes are dipping in the NE and SW quadrants, 43 of them trend to the ENE and to the SSW. Only 14 planes (7%) trend NW, sub-parallel to the fault. The distribution of inclinations covers the entire range, except for the planes that trend with the fault. Of these, 9 are sub-vertical and 3 are sub-horizontal. The slicknsides were imaged as lines in space but the senses of motion have not yet been determined. They are low-angle 3D lines with no consistent trend indicating mostly horizontal motion. The mechanical origin of this set of surfaces needs a further verification. While they are not consistent with Riedel shear geometry, the characteristics of the surfaces show that they accommodate a small amount of rotation that may reflect simple shear. This requires a relatively homogeneous stress field. As this is not necessarily the case, they might be of dynamic origin, reflecting the heterogeneity of a process-zone stress field ahead of the earthquake rupture. Finally, the dynamic reduction of normal stress during the passage of a rupture can control the formation of this geometry, but requires that the slicknsides (to be verified) will demonstrate a tensile regime.

Bookmarks Related papers MentionsView impact

Research paper thumbnail of Binocular Stereo

Bookmarks Related papers MentionsView impact

Research paper thumbnail of A perceptual organization approach for figure completion, binocular and multiple-view stereo and machine learning using tensor voting

Bookmarks Related papers MentionsView impact

Research paper thumbnail of Real-time video-based reconstruction of urban environments

Proceedings of 3DARCH: 3D Virtual Reconstruction and Visualization of Complex Architectures, Jul 1, 2007

1University of North Carolina, Chapel Hill, NC, USA {mordohai, jmf, bclipp, gallup, pmerrell, sal... more 1University of North Carolina, Chapel Hill, NC, USA {mordohai, jmf, bclipp, gallup, pmerrell, salmi, taltonb, herman, welch, marc}@ cs. unc. edu 2 University of Kentucky, Lexington, KY, USA {amir, engels, qingxiong. yang, stewe}@ vis. uky. edu,{lwangd, ryang, dnister}@ cs. uky. edu KEY WORDS: computer vision, 3D model, stereo reconstruction, large-scale reconstruction, real time processing, geo-registration

Bookmarks Related papers MentionsView impact

Research paper thumbnail of Binocular and Multiple View Stereo using Tensor Voting

Bookmarks Related papers MentionsView impact

Research paper thumbnail of Inference of Segmented Overlapping Surfaces From Binocular and Multiple-View Stereo

We present an integrated approach,to the derivation of scene descriptions from two or more stereo... more We present an integrated approach,to the derivation of scene descriptions from two or more stereo images, where evaluation of feature correspondences and grouping into perceptual structures are addressed,within the same,framework.,Special attention is given to the development of a methodology with general applicability. In order to handle the issues of noise, lack of image features, surface discontinuities and half occluded

Bookmarks Related papers MentionsView impact

Research paper thumbnail of Dimensionality Estimation and Manifold Learning using Tensor Voting

We address instance-based learning from a perceptual organization standpoint us- ing tensor votin... more We address instance-based learning from a perceptual organization standpoint us- ing tensor voting. The goal of instance-based learning is to learn the underlying relationships between observations, which are points in an N-D continuous space, under the assumption that they lie in a limited part of the N-D space, typically a manifold, the dimensionality of which is an indication of the

Bookmarks Related papers MentionsView impact

Research paper thumbnail of Unsupervised Dimensionality Estimation and Manifold Learning in high-dimensional Spaces by Tensor Voting

International Joint Conference on Artificial Intelligence, 2005

We address dimensionality estimation and nonlin- ear manifold inference starting from point input... more We address dimensionality estimation and nonlin- ear manifold inference starting from point inputs in high dimensional spaces using tensor voting. The proposed method operates locally in neigh- borhoods and does not involve any global com- putations. It is based on information propagation among neighboring points implemented as a voting process. Unlike other local approaches for man- ifold learning, the quantity

Bookmarks Related papers MentionsView impact

Research paper thumbnail of Object Detection from Large-Scale 3D Datasets Using Bottom-Up and Top-Down Descriptors

Lecture Notes in Computer Science, 2008

ABSTRACT We propose an approach for detecting objects in large-scale range datasets that combines... more ABSTRACT We propose an approach for detecting objects in large-scale range datasets that combines bottom-up and top-down processes. In the bottom-up stage, fast-to-compute local descriptors are used to detect potential target objects. The object hypotheses are verified after alignment in a top-down stage using global descriptors that capture larger scale structure information. We have found that the combination of spin images and Extended Gaussian Images, as local and global descriptors respectively, provides a good trade-off between efficiency and accuracy. We present results on real outdoors scenes containing millions of scanned points and hundreds of targets. Our results compare favorably to the state of the art by being applicable to much larger scenes captured under less controlled conditions, by being able to detect object classes and not specific instances, and by being able to align the query with the best matching model accurately, thus obtaining precise segmentation.

Bookmarks Related papers MentionsView impact

Research paper thumbnail of A hybrid control approach to the Next-Best-View problem using stereo vision

2013 IEEE International Conference on Robotics and Automation, 2013

Bookmarks Related papers MentionsView impact

Research paper thumbnail of Tensor Voting: A Perceptual Organization Approach to Computer Vision and Machine Learning

Synthesis Lectures on Image, Video, and Multimedia Processing, 2006

Bookmarks Related papers MentionsView impact

Research paper thumbnail of Multi-View Stereo via Graph Cuts on the Dual of an Adaptive Tetrahedral Mesh

2007 IEEE 11th International Conference on Computer Vision, 2007

Bookmarks Related papers MentionsView impact

Research paper thumbnail of Correspondence estimation for non-rigid point clouds with automatic part discovery

The Visual Computer, 2015

Bookmarks Related papers MentionsView impact

Research paper thumbnail of Classification of Vehicle Parts in Unstructured 3D Point Clouds

2014 2nd International Conference on 3D Vision, 2014

Bookmarks Related papers MentionsView impact