M.Sc. Anas Al-Nuaimi

 
M.Sc. Anas Al-Nuaimi

M.Sc. Anas Al-Nuaimi

Telefon: 
Fax: +49 (89) 289 - 23523
Raum: 0407

Technische Universität München
Lehrstuhl für Medientechnik
80290 München

 
rendered: 2017-11-22 06:19:43

Biography

I studied Electrical and Computer Engineering at the Hashemite University in Jordan majoring in Telecommunications. After that I pursued a master's degree at TUM's international M.Sc. degree program in Communications Engineering (MSCE) majoring in Communications Systems. In my master's thesis on the topic of "Rapid Feature Matching" I demonstrated the potential behind using visual localization systems for which I was awarded the E-ON future award for outstanding thesis work. During my master's study I also obtained an honors degree in Technology Management from the Center for Digital Technology and Management (CDTM).

I joined the Chair of Media Technology as member of the research staff and doctoral student in April 2010. Initially I worked in a collaborative project with Docomo Eurolabs on the topic of video synchronization of multi-perspective video. In the follow-up project, CoopMedia 3D, I researched techniques to enrich RGB images with depth data for more accuarte location retrieval in GPS-denied indoor environment with sparse appearance features. Late 2013 I joined the exciting NavVis project which later became a spin-off. There I enjoyed collaborating with a very smart and dynamic team on various topics of computer vision and 3D processing. My expertise lies in the area of automatic alignment of 3D point clouds and 3D sensing.


In 2016 I spent a month at the University of Sao Paulo working on 3D LIDAR scan matching. I also got the chance to explore a small part of Brazil. Some of my impressions you can read on wordpress.

Enhanced Depth Estimation using a Combination of Structured Light Sensing and Stereo Reconstruction (using the Google Tango/Peanut) [1]

We present an approach for joint stereo-reconstruction and structured-light sensing for accuracte 3D scanning of indoor scenes using a smartphone. For this work we use the Google Project Tango device.

We use the Google Tango device.
It can capture RGB images as well as depth images in real time.
Image source: Youtube (click on image)
Sample scene reconstructed using stereo and structured light (Tango's sensor).
Stereo excels at well-textured surfaces but fails on non-textured surfaces. Structured-light fails on glossy surfaces. Also the it cannot depth surfaces beyond a certain depth.
A simple fusion shows the potential behind joint reconstruction.
Naive fusion is undesirable as it does not involve regularization and simply selects one of the depth values if both are available instead of making a more informed decision.

Approach

The global energy function a la Kolmogrov and Zabih [2] is adapted to involve stereo and structured light while considering the sensor characterisitcs of the structured-light sensor (see the paper [1]).

The fusion is done in an iterative process starting with the low resolution of the structured-light sensor and ending with a resolution matching that of the RGB sensor.

Results

Compared to simple stereo reconstruction and plain structured-light (Tango sensor only) the fused depth map is much denser covering regions that can not be sensed by stereo or structured-light. What is not seen is that the fused depth map has 16 times the resolution (4x increase in both dimensions).
The adapted energy function with improved smoothness constraint exploits prior depth knowledge from the structured-light sensor to compute depth gradients which are used to improve the reconstruction of slanted surfaces (walls not perpendicular to the imaging plane)
The iterative upsampling is shown to decrease the depth measurement error as a consequence of the resolution increase which leads to more accurate disparity computation. Especially distant regions (a, b and d) benfit from this accuracy improvement.
Compared to standard stereo, the required computational time is reduced by 90%.

References

[1]
A. Wittmann, A. Al-Nuaimi, E. Steinbach, G. Schroth,
Enhanced Depth Estimation using a Combination of Structured Light Sensing and Stereo Reconstruction (using the Google Tango/Peanut),
11th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications., Rome, Italy, Februar 2016.
Download BibTeX entryDownload fulltext PDF

[2] Kolmogorov, Vladimir, Pascal Monasse, and Pauline Tan. "Kolmogorov and Zabih's Graph Cuts Stereo Matching Algorithm." Image Processing On Line 4 (2014): 220-251

Indoor Location Retrieval with kinFu and Large-Scale Point Clouds [1]

We show that indoor location retrieval can be posed as a partial shape matching problem of Kinect
Fusion (KinFu) query scans in large-scale reference (target) indoor point clouds.

Obtaining query scan with a Kinect-like sensor + KinectFusion
Matching the query to a large-scale point cloud allows retrieving the location and pose accurately

Used Retrieval System

A local-shape feature based retrieval system is used. Descriptors are computed for keypoints detected in the reference cloud and query cloud. The descriptors are matched and RANSAC is used to estimate the transformation hypothesis robustly. Finally, ICP refines the transformation estimate resulting in very accurate pose estimation.

Problem and Solution

Raw depth data exhibits bending on flat surfaces which increases with increasing depth.
The depth data bending leads to bent KinFu scans. When scanning a scene from a closer distance (2) the bending is reduced.
We use the CLAMS calibration technique [2]. The sensor calibration reduces the bending in the raw data.
The sensor calibration leads to substantially reduced bending in our KinFu scans.

Results

In the paper we show this calibration is critical to good retrieval performance. The video below shows how the retrieval is performed.

Datasets

The target clouds can be downloaded from here.

References

[1]
A. Al-Nuaimi, M. Piccolrovazzi, S. Gedikli, E. Steinbach, G. Schroth,
Indoor Location Retrieval using Shape Matching of KinectFusion Scans to Large-Scale Indoor Point Clouds,
Workshop on 3D Object Retrieval co-located with Eurographics 2015, Zurich, Switzerland, Mai 2015.
Download BibTeX entryDOIDownload fulltext PDF

[2] Alex Teichman, Stephen Miller, and Sebastian Thrun.  Unsupervised intrinsic calibration of depth sensors via SLAM. Robotics: Science and Systems (RSS), 2013.

Video Synchronization

Within the Cooperative Mobile Media Project we work on multi-perspective recordings in UGC scenarios. Such recordings capture interesting events such as musical performances, sports events or street events. The problem with UGC multi-perspective recordings is that the uploaded videos are typically not temporally synchronized. At the LMT, we have developed a method to temporally synchronize multi-perspective events by cross-correlating bitrate profiles[1]. We have also developed a method for allowing robust cross-correlation of signals to handle occasional signal mismatches [2] as shown in the figure below.

 

Recently we presented an extension of the ConCor idea based on proven concepts from the iterative closest point algorithm to allow highly robust and confident synchronization as well as enable joint audio-video synchronization and hence exploit all modalities of the videos[3].

References

[1]
A. Al-Nuaimi, B. Cizmeci, F. Schweiger, R. Katz, S. Taifour, E. Steinbach, M. Fahrmair,
ConCor+: Robust and Confident Video Synchronization using Consensus-based Cross-Correlation,
IEEE International Workshop on Multimedia Signal Processing, Banff, AB, Canada, September 2012.
Download BibTeX entryDOIDownload fulltext PDF
[2]
F. Schweiger, G. Schroth, M. Eichhorn, E. Steinbach, M. Fahrmair,
Consensus-based Cross-correlation,
ACM Multimedia, Scottsdale, AZ, November 2011.
Download BibTeX entryDOIDownload fulltext PDF
[3]
G. Schroth, F. Schweiger, M. Eichhorn, E. Steinbach, M. Fahrmair, W. Kellerer,
Video Synchronization using Bit Rate Profiles ,
IEEE International Conference on Image Processing (ICIP), Hong Kong, September 2010.
Download BibTeX entryDOIDownload fulltext PDF

Journal Papers

[5]
A. Al-Nuaimi, S. Hilsenbeck, A. Garcea, E. Steinbach,
6DOF Decoupled Roto-Translation Alignment of Large-Scale Indoor Point Clouds,
Computer Vision and Image Understanding (CVIU), Special Issue on Large-Scale 3D Modeling of Urban Indoor or Outdoor Scenes from Images and Range Scans, September 2016.
Download BibTeX entryDOI
[4]
W. Lopes, A. Al-Nuaimi, C. Lopes,
Geometric-Algebra LMS Adaptive Filter and its Application to Rotation Estimation,
IEEE Signal Processing Letters, vol. 23, no. 6, pp. 858 - 862, Juni 2016.
Download BibTeX entryDOI
[3]
X. Xu, B. Cizmeci, A. Al-Nuaimi, E. Steinbach,
Point Cloud-based Model-mediated Teleoperation with Dynamic and Perception-based Model Updating,
IEEE Transactions on Instrumentation and Measurement , vol. 63, no. 11, pp. 2558 - 2569 , 2014.
Download BibTeX entryDOIDownload fulltext PDF
[2]
F. Schweiger, G. Schroth, M. Eichhorn, A. Al-Nuaimi, B. Cizmeci, M. Fahrmair, E. Steinbach,
Fully Automatic and Frame-accurate Video Synchronization using Bitrate Sequences,
IEEE Transactions on Multimedia, vol. 15, no. 1, pp. 1-14, 2013.
Download BibTeX entryDOIDownload fulltext PDF
[1]
G. Schroth, R. Huitl, D. Chen, M. Abu-Alqumsan, A. Al-Nuaimi, E. Steinbach,
Mobile Visual Location Recognition,
IEEE Signal Processing Magazine, Special Issue on Mobile Media Search, vol. 28, no. 4, pp. 77-89, July 2011.
Download BibTeX entryDOIDownload fulltext PDF

Papers in Conference Proceedings

[10]
A. Al-Nuaimi, W. Lopes, P. Zeller, A. Garcea, C. Lopes, E. Steinbach,
Analyzing LiDAR Scan Skewing and its Impact on Scan Matching ,
Seventh Int. Conf. on Indoor Positioning and Indoor Navigation, Madrid, Spain, Oktober 2016.
Download BibTeX entry
[9]
A. Al-Nuaimi, W. Lopes, E. Steinbach, C. Lopes,
6DOF Point Cloud Alignment using Geometric Algebra-based Adaptive Filtering,
IEEE Winter Conference on Applications of Computer Vision (WACV2016), Lake Placid, NY, USA, März 2016.
Download BibTeX entryDownload fulltext PDF
[8]
A. Wittmann, A. Al-Nuaimi, E. Steinbach, G. Schroth,
Enhanced Depth Estimation using a Combination of Structured Light Sensing and Stereo Reconstruction (using the Google Tango/Peanut),
11th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications., Rome, Italy, Februar 2016.
Download BibTeX entryDownload fulltext PDF
[7]
A. Al-Nuaimi, M. Piccolrovazzi, S. Gedikli, E. Steinbach, G. Schroth,
Indoor Location Retrieval using Shape Matching of KinectFusion Scans to Large-Scale Indoor Point Clouds,
Workshop on 3D Object Retrieval co-located with Eurographics 2015, Zurich, Switzerland, Mai 2015.
Download BibTeX entryDOIDownload fulltext PDF
[6]
J. Chao, A. Al-Nuaimi, G. Schroth, E. Steinbach,
Performance Comparison of Various Feature Detector-descriptor Combinations for Content-based Image Retrieval with JPEG-encoded Query Images,
IEEE International Workshop on Multimedia Signal Processing (MMSP), Pula, Sardinia, Italy, Oktober 2013.
Download BibTeX entryDOIDownload fulltext PDF
[5]
S. Apostu, A. Al-Nuaimi, E. Steinbach, M. Fahrmair, X. Song, A. Möller,
Towards the design of an intuitive Multi-View Video navigation interface based on spatial information,
15th International Conference on Human-Computer Interaction with Mobile Devices and Services (mobileHCI 2013), Munich, Germany, August 2013.
Download BibTeX entryDOIDownload fulltext PDF
[4]
A. Al-Nuaimi, R. Huitl, S. Taifour, S. Sarin, X. Song, Y. Gu, E. Steinbach, M. Fahrmair,
Towards location recognition using range images,
2013 IEEE International Conference on Multimedia and Expo (ICME2013), Workshop on Hot Topics in 3D (Hot3D), San José, USA, July 2013.
Download BibTeX entryDOIDownload fulltext PDF
[3]
A. Al-Nuaimi, B. Cizmeci, F. Schweiger, R. Katz, S. Taifour, E. Steinbach, M. Fahrmair,
ConCor+: Robust and Confident Video Synchronization using Consensus-based Cross-Correlation,
IEEE International Workshop on Multimedia Signal Processing, Banff, AB, Canada, September 2012.
Download BibTeX entryDOIDownload fulltext PDF
[2]
H. Zhang, X. Gu, A. Al-Nuaimi, M. Fahrmair, R. Ishibashi,
Seamless and Efficient Stream Switching of Multi- Perspective Videos,
International Packet Video Workshop 2012, Munich, Germany, Mai 2012.
Download BibTeX entryDOI
[1]
G. Schroth, A. Al-Nuaimi, R. Huitl, F. Schweiger, E. Steinbach,
Rapid Image Retrieval for Mobile Location Recognition,
IEEE International Conference on Acoustics, Speech and Signal Processing, Prague, Czech Republik, Mai 2011.
Download BibTeX entryDOIDownload fulltext PDF

Präsentationen

[1]
A. Al-Nuaimi,
Vollautomatische und bildgenaue Synchronisierung von Videosequenzen,
ITG-Fachausschuss 3.2, Januar 2015.
Download BibTeX entryDownload fulltext PDF
 
 Drucken