MVT - 1 Video Transmission over Wireless Networks
NICTA Project - Dr. Jing Chen (Jing.Chen@nicta.com.au)
Sending video sequence over wireless network where retransmission is infeasible or too expensive is a hot topic for video communication over wireless networks. Cross-Layer coding has been considered as a promising solution to offer error resilience in video communication especially under varying and limited channel bandwidth situation. In this project, you will study the architecture of the cross-layer video coding, investigate kinds of existing cross-layer coding technologies for a wireless network, and finally setup a simulation platform to evaluate the performance of your schemes. After this project, you will gain some solid knowledge and experience at video coding such as MPEG4 and H.264 etc, TCP/IP, and the protocol of networking player of a wireless network.
This project can also build up your programming skills in C/C++. An expected outcome of the research is to gain some initial results at video transmission over wireless mesh networks.
For further information, please contact the supervisor or consider the following:
Yan Wang, "Multiple Description Coding for Video Delivery", PROCEEDINGS OF THE IEEE, VOL. 93, NO. 1, JANUARY 2005
V. K. Goyal, "Multiple description coding: Compression meets the network," IEEE Signal Processing Mag., vol. 18, pp. 74-93, Sep. 2001.
MVT - 2 Objects Classification and Event Detection in Surveillance Video
NICTA Project - Associate Professor Jian Zhang (Jian.zhang@nicta.com.au), Getian Ye (Getian.ye@nicta.com.au) and Sijun Lu (Sijun.lu@nicta.com.au)
Object classification is a further step to object detection and tracking in understanding of video. Without object classification, we can only know something is moving in the video sequences, but don't know what it is. With the information from object tracking and classification, it may become possible to understand the activities and events happening in the video sequences. The nature of the work is software implementation to investigate the state-of-art object classification technologies for categorizing moving vehicles or human detected from the surveillance videos. The knowledge you learned from the courses of computer vision, neural network, pattern recognition and machine learning can all find their utilization in this project. Your programming skills will also be developed as you will develop a demo in software. This project is an absolute opportunity for a ToR student to test his/her research capabilities for the PhD study.
The novelty lies on the new approach for objects classification and statistical feature clustering. You will access the current state-of-art image/video processing in-house-made tools. Your work will contribute to NICTA's Strategic Project - Smart Transport and Road (STaR). An expected outcome of the research is algorithm software implementation and some test results.
For further information, please contact the supervisors or consider the following:
T. Kanade, R. Collins, A. Lipton, P. Burt, and L. Wixson, "Advances in Cooperative Multi-Sensor Video Surveillance",Darpa Image Understanding Workshop, Morgan Kaufmann, November, 1998, pp. 3-24
R. Collins, A. Lipton, H. Fujiyoshi, and T. Kanade, "Algorithms for cooperative multisensor surveillance", Proceedings of the IEEE, Vol. 89, No. 10, October, 2001, pp. 1456 - 1477.
Wei Niu, Jiao Long, Dan Han and Yuan-Fang Wang, "Human Activity Detection and Recognition for Video Surveillance", ICME 2004
Surendra Gupte, Osama Masoud, Robert F. K. Martin, and Nikolaos P. Papanikolopoulos, "Detection and Classification of Vehicles", IEEE Transactions on Intelligent Transportation Systems, VOL. 3, NO. 1, December 2002
MVT - 3 A Real-time and Robust Object Tracking System for the Video Surveillance.
NICTA Project - Associate Professor Jian Zhang (Jian.zhang@nicta.com.au), Getian Ye (Getian.ye@nicta.com.au) and Sijun Lu (Sijun.lu@nicta.com.au)
The advance of fast and affordable computing device has provided the power to track moving objects from a video camera in real-time. Many algorithms have been developed for object tracking in the past, using various image/video features, such edge, color, spatial-temporal information, or using Kalman filtering and hypothesis testing techniques. The nature of this work is to implement the existing object tracking techniques and choose one suitable for the video surveillance domain and the real-time purpose. Research challenge of this task is to achieve robust tracking under the disturbing environment such as occlusion and lighting changes. The knowledge you learned from the computer and engineering courses, such as signal processing, graphic theory and computer vision will gain plenty practice from this project. Your programming skills will also be developed as you will develop a demo in software. This project is an absolute opportunity for a ToR student to test his/her research capabilities for the PhD study.
The novelty lies on the new approach to tracking objects without implementing background subtraction. You will access the current state-of-art image/video processing in-house-made tools. Your work will contribute to NICTA's Strategic Project - Smart Transport and Road (STaR). An expected outcome of the research is algorithm software implementation and some test results.
For further information, please contact the supervisors or consider the following:
T. Kanade, R. Collins, A. Lipton, P. Burt, and L. Wixson, "Advances in Cooperative Multi-Sensor Video Surveillance",Darpa Image Understanding Workshop, Morgan Kaufmann, November, 1998, pp. 3-24
R. Collins, A. Lipton, H. Fujiyoshi, and T. Kanade, "Algorithms for cooperative multisensor surveillance", Proceedings of the IEEE, Vol. 89, No. 10, October, 2001, pp. 1456 - 1477.
Tao Xiong and Christian Debrunner, "Stochastic Car Tracking With Line- and Color-Based Features", IEEE Transactions on Intelligent Transportation Systems, VOL. 5, NO. 4, December 2004
Yue Zhou and Hai Tao, "A Background Layer Model for Object Tracking through Occlusion", ICCV 2003
MVT - 4 Content-based Image Retrieval with Support Vector Machines
NICTA Project - Associate Professor Jian Zhang (Jian.zhang@nicta.com.au), Getian Ye (Getian.ye@nicta.com.au) and Sijun Lu (Sijun.lu@nicta.com.au)
Feature selection has been a preliminary step to pre-process high dimensional data for mining, indexing and visualization. Generally, functionality of feature selection includes two major components 1) dimension reduction - effectively reducing size of raw feature size, 2) optimal feature subset selection. In this project, we mainly focus on large multimedia data. An advance feature selection scheme will be developed to facilitate efficient and effective classification of large multimedia dataset and compared with state of art approaches. The nature of the work focuses on algorithm software development. It will be carried out in multimedia & video communication research group located in one of National ICT Australia (NICTA) laboratories - Kensington Lab (L5 building). This project is an absolute opportunity for a ToR student to test his/her research capability for the PhD study
The novelty lies on the new approach for image Castigation based on feature extraction. You will access the current state-of-art image/video processing in-house-made tools. Your work will contribute to NICTA's project - Video analysis and content management for surveillance (VACMS). An expected outcome of the research is algorithm software implementation and some test results.
For further information, please contact the supervisors or consider the following:
S.F. Chang, W. Chen, H.J. Meng, H. Sundaram, D. Zhong, "A Fully Automated Content-Based Video Search Engine Supporting Spatiotemporal Queries" IEEE Transaction on Circuits and Systems For Video Technology, Vol 8, No. 5, September 1998.
R.C. Gonzalez, R.E Woods, "Digital Image Processing" Second Edition.
Y.X Chen & J.Z, Wang, "Image Categorization by Learning and Reasoning with Region; The Journal of Machine Learning Research, VOL. 5, DECEMBER 2004
C. Chang, C. Lin. A Comparison of Methods for Multi-class Support Vector Machines, http://www.csie.ntu.edu.tw/~cjlin
MVT - 5 Pedestrian Detection Using a Cascade of Boosted Classifiers
NICTA Project - Associate Professor Jian Zhang (Jian.zhang@nicta.com.au) and Sijun Lu (Sijun.lu@nicta.com.au)
This project aims provide an implementation, verification and evaluation of a pedestrian detection framework proposed by Viola-Jones which is fast and robust under varying conditions by incorporating both motion and appearance in training a cascade classifiers. It is hoped that this project will serve as a foundation for other applications such as vehicle and cyclist detection; and incorporated with other image processing techniques such as tracking which would improve robustness and computational efficiency. The nature of this work will focus on software implementation based on state-of-art international conference papers.
The novelty lies on the new approach for image categorization based on feature extraction. You will access the current state-of-art image/video processing in-house-made tools. Your work will contribute to NICTA's project - Video analysis and content management for surveillance (VACMS). An expected outcome of the research is algorithm software implementation and some test results.
For further information, please contact the supervisors or consider the following:
Paul Viola, Michael J. Jones, Daniel Snow. "Detecting Pedestrians using Patterns of Motion and Appearance", ICCV 2003. November 27
http://www.merl.com/papers/docs/TR2003-90.pdf
MVT - 6 Video coding and transmission over wireless networks
NICTA Project - Associate Professor Jian Zhang (Jian.zhang@nicta.com.au) and Raymond Leung (raymond.leung@nictta.com.au)
Wireless video/image communication will bring huge multimedia applications. As a strong demand from mobile and PDA industry, the research and R/D are moving fast. The video coding and its communication protocol have been developed by the MPEG/ITU-T standard organizations and Internet Engineering Task Force (IETF). This project is to investigate the practical error resilience scheme for robust video transmission over mesh wireless networks. It provides a significant opportunity for students to understand image/video coding algorithm and error resilience for video communication. After this project, you will develop some solid knowledge in the area of video communications. The nature of this work focuses on software implementation based on existing algorithms. This project will require some C/C++ coding skills and some courses training including multimedia technology, advanced Math, software development and image processing. This research could be extended to a postgraduate study.
The novelty lies on the new approach for investigating a new scheme to maximize the cross layer performance including source and channel coding and transmission layers. You will access the current state-of-art image/video coding tools. Your work will contribute to NICTA's project - Video analysis and content management for surveillance (VACMS) and Smart Road and Transportation (STaR) project. An expected outcome of the research is algorithm software implementation and some test results.
For further information, please contact the supervisors or consider the following:
Ghanbari, M. Video coding: an introduction to standard codecs. London : Institution of Electrical Engineers, c1999.
WIKIPEDIA. (2006). "Video Compression", http://en.wikipedia.org/wiki/Video_compression
Ian F. A. et al. Wireless mesh networks: a survey. 2005. www.sciencedirect.com
Wang et al. Multiple Description Coding for Video Delivery. In Preceedings of the IEEE, VL. 93, NO. 1, JANUARY 2005.
Information Sciences Institute, Official ns-2 website. "The Network Simulator - ns-2". http://www.isi.edu/nsnam/ns/
Worcester Polytechnic Institute. "NS by Example". http://nile.wpi.edu/NS/
[Top of Page]
