The Interactive Systems Design Lab within the EE Department's Applied Signal and Image Processing Division
Principal Investigator: Professors Les Atlas and John Sahr
Sponsor: The Office of Naval Research Signal Analysis Program
Abstract: The theoretical link between a discrete-time sequence and its discrete-time/discrete-frequency representation has heretofore been established via a uniform sampling of their continuous-time counterparts. In this research, we propose to provide a direct link between the two which we establish using the concepts of operator theory. We so far see that many similarities, but also some important differences, exist between the results of the continuous-time operator approach and our discrete one. The differences between the continuous distributions and discrete ones may not be the simple sampling relationship which has so often been assumed. Through basic matrix operations, discrete-time/discrete-frequency distributions can be generated using our operators, and we show that: a) key properties like positivity are much easier to formulate and solve in the discrete case, and b) while proper quadratic distributions are not possible using the Fourier transform, they do indeed exist for other transforms. We wish to develop these results and extend them to our previously-developed hidden Markov model-based automatic classification system, with verification in high priority Navy applications.
References:
J. W. Pitton, and L. E. Atlas, "Discrete-time Implementation of the Cone-Kernel Time-Frequency Representation," IEEE Transactions on Signal Processing, Vol. 43, No. 8, Aug. 1995, pp. 1996-8.
J. McLoughlin, J. Droppo, L. Atlas, "Class-Dependent, Discrete Time-Frequency Distributions via Operator Theory," Proc. ICASSP `97, April 21-24, 1997, Munich, Germany.
http://isdl.ee.washington.edu/isdl/isdl.html
Computational Intelligent Applications Laboratory
Principal Investigator: Mohamed A. El-Sharkawi
Sponsor: Southern California Edison Company, Bonneville Power Administration
Abstract: Switching transients occurring in power systems are often due to circuit breaker (CB) switching. CBs are not designed to close or open at the time of minimum stress (zero voltage for closing and zero current for opening). For example, when a circuit breaker switches a capacitor bank, large and damaging voltage and current transients may occur. Depending on the switching instance, the bus voltage may collapse momentarily to zero then oscillate at high frequencies with high magnitudes. When the circuit breaker of a capacitor bank opens, an arc is often produced which causes transients in current and voltage. To avoid these transients, CBs should ideally be closed at zero voltage and opened at zero current. This is achievable by the Adaptive Sequential Controller (ASC) developed in the Energy Laboratory of the University of Washington. The ASC has the following general features:
Its power electronic circuit allows each phase of a circuit breaker to close at or near zero voltage. The controller can be programmed to trigger the magnetic mechanism at a precise time, taking into account the speed of the breaker's operating mechanism. The triggering circuit is activated so that when the circuit breaker contacts close, the voltage across the breaker is at or near zero. The same operating principal is used to open the CB at the point of minimum stress. When the breaker speed drifts due to weather or normal wear, the ASC automatically compensates for the change without human intervention.
The ASC is a "Fail-Safe" device. If the ASC fails, it will automatically disconnect from the circuit breaker operating mechanism so that the breaker can operate manually or remotely in a non-sequential manner.
The ASC can be used with any single pole breaker, or modern multi-phase breakers with independent phase magnetic operating mechanisms (solenoids). For multi-phase, mechanically-ganged breakers, where only one solenoid is employed, the ASC could be used after additional operating mechanisms are retrofitted to the other phases. At this stage of development, the ASC is not designed to operate as described during faults. Fault interruption will be done by the normal fault clearing mechanism of the circuit breaker.
References:
M. A. El-Sharkawi, A. Szofran, T. Huang, G. Andexler, M. Dong, S.S. Venkata, A. Rodriguez, N. Butler, A. Van Leuven and D. Smith, "Development and Field Testing of an Adaptive Flicker Controller for 15-kV Systems," IEEE Transaction on Power Delivery, April 1995, pp 1025-1030.
Patents:
"High Voltage Solid State Switching Circuit," US Patent No. 5,180,963, Jan. 19, 1993. The co-inventors are George Andexler and Lee Silberkliet.
"Adaptive Sequential Controller for Power System Circuit Breakers," US Patent No. 5,361,184, Nov. 1, 1994. The co-inventors are Jian Xing, Alanso Rodriguez and Nickolas Butler.
"Adaptive Sequential Controller with Minimum Switching Energy," The co-inventors are Jian Xing, Alanso Rodriguez and Nickolas Butler. Patent approved.
http://cialab.ee.washington.edu/home.html-ssi
Computational Intelligence Applications Laboratory
Principal Investigator: Mohamed A. El-Sharkawi and Robert J. Marks II
Sponsor: National Science Foundation and Electric Power Research Institute
Abstract: One of the most difficult problems in the operation of large synchronous turbine-generators is the detection of shorted turns in the DC-field of the rotor. Not only is the existence of a shorted turn in the field winding hard to detect, its correction may result in an expenditure of several hundred thousand dollars when including the cost of replacing the lost power generation with more expensive sources such as large nuclear powered machines. Unfortunately, this expense is incurred even in the case of a wrong diagnosis. This is because the major expense results from the disassembly and assembly of the machine and in the added cost of alternative production. Proper localization, and more important, accurate determination of the actual existence of a shorted-turn, is therefore essential to avoid huge unnecessary monetary losses. A general solution to this problem has so far remained elusive. In this project a twin signal signature sensing is used to monitor, detect, and localize shorts in power system equipment with windings, including rotors, transformers, motors and large synchronous turbine-generators. There has, to date, been no effective way to do so. The most obvious approach, time domain reflectometry, fails due to the reactive coupling in the windings. Twin signal signature sensing of shorts results from identical signals being simultaneously injected in both sides of the windings. The transmitted signals are difference to obtain the signature signal of the device. Through the monitoring of the evolution of the signature signals, development of winding shorts can be diagnosed through the process of novelty detection. Windings with shorts previously fingerprinted can be subjected to tests to localize the shorts. The standard layered perceptron neural network appears ideal to make these decisions. Preliminary work, performed on both downed and rotating loaded rotors, has been quite promising in demonstrating the effectiveness of the twin signal signature sensing approach to winding short evolution monitoring.
References:
M.A. El-Sharkawi and R.J. Marks II, "What role can neural networks play in power system engineering," IEEE Power Engineering Review, February 1994, pp. 14-16.
M.A. El-Sharkawi, R.J. Marks II, S.Oh, S.J. Huang, I. Kerszenbaum and A. Rodriguez, "Localization of Winding Shorts Using Fuzzified Neural Networks," IEEE Transactions on Energy Conversion, Vol. 10, No.1, March 1995, pp.147-155.
R.J. Streifel, R.J. Marks II, M.A. El-Sharkawi and I. Kerszenbaum, "Twin Signal Sensing: Application to Shorted Winding Monitoring, Detection and Localization," Applications of Neural Networks in Environment, Energy and Health, P.E. Keller, S.Hashem, L.J. Kangas and R.T. Kouzes, Editors, (World Scientific, Singapore, 1995), pp. 133-134
http://cialab.ee.washington.edu/home.html-ssi
Computational Intelligent Applications Laboratory
Principal Investigator: Mohamed A. El-Sharkawi and Robert J. Marks II
Sponsor: Southern California Edison Company
Abstract: The rapid growth of information processing technology in recent years has created new opportunities for implementation of a new class of computationally intensive paradigms that mimic human intelligence. Collectively, these disciplines are referred to as computational intelligence (CI). The field includes artificial neural networks, evolutionary computation and certain fuzzy systems. Together with expert systems, the most successful application of conventional artificial intelligence, CI has found use in the analysis, diagnosis and control of numerous systems. Intelligent systems are those utilizing computational intelligent or expert system algorithms.
A large number of transmission and distribution problems share a set of common characteristics that make them candidates for intelligent system solutions. A large scale transmission and distribution system with its varying equipment, interutility connections, operational strategies, and safety concerns, poses demanding computational issues in planning, control, diagnosis and protection. A number of intelligent solutions have been proposed for solving certain transmission and distribution problems. The initial results of these applications are very promising. As a result, many utilities are currently studying, developing, and implementing intelligent systems tools.
The proposed areas of research for this project are divided into two broad areas:
1. Equipment Diagnosis and Control; and
2. System Diagnosis and Control.
In this phase, a detailed review of the state of the art of intelligent system technologies and transmission and distribution applications will be made. Selected intelligent system techniques applicable to transmission and distribution problems will be identified. In addition, integrated computationally intelligent tools will be investigated for use in transmission and distribution applications.
http://cialab.ee.washington.edu/home.html-ssi
Computational Intelligent Applications Laboratory
Principal Investigators: M. A. El-Sharkawi, Robert J. Marks II and Russell Reed
Abstract: In the past decades, Energy Management Systems (EMS) have provided electric utilities with On-line Static Security Assessment tools to provide coverage for steady state operating conditions of the system. Currently, the power system industry's attention is focused upon providing On-line Dynamic Security Assessment. With the accelerated growth of computing power and recently introduced analytical methods, the possibility of have On-line Dynamic Security Assessment (DSA) is becoming a reality.
Dynamic Security Assessment's function is to determine which contingencies may cause power system limit violations or system instability. A brute force approach would be to conduct detail stability analysis for each credible contingency followed by a check for violations. However this approach is slow. The existing DSA practice is to carry out off-line studies with the most stressed operating conditions in deriving guidelines for the operators. However, during actual system operations, conditions seldom match the situations studied off-line. Consequently, the guidelines and limits produced are too conservative with significant financial consequences to the utilities. A more effective approach is to assess only those contingencies likely to cause dynamic violations in real-time.
There are a number of areas in the pre and post processing of DSA whose performance need enhancements. In post processing, significant improvements in performance may be realized if a package can screen the immense quantities of output data from a stability run, perform analysis and make high speed decisions, and finally control the computational process in reducing the number of studies required. Neural networks and artificial intelligence may hold the key to substantially enhancing the performance and streamlining some of the DSA tasks. Artificial neural networks (ANN) have been studied for many years with the hope of achieving human-like performance in solving certain problems in speech and image processing. In the recent years, there has been a resurgence in the field of neural networks due to the introduction of new network topologies, training algorithms and VLSI techniques. The potential benefits of neural networks; such as parallel distributed processing, high computation rates, fault tolerance, and adaptive capability have lured reseachers from fields such as power systems, controls, signal processing and robotics to seek neural network solutions to some of their more challenging problems. The purpose of this research is to evaluate the applications of neural networks to Dynamic Security Assessment (DSA). The project gives a working design for an on-line DSA from a Canadian utility to provide understanding into the various problems associated with On-line DSA.
References:
Y. Mansour, E. Vaahedi, A.Y. Chang, B.R. Corns, J. Tamby and M. A. El-Sharkawi, "Large Scale Dynamic Security Screening and Ranking Using Neural Networks," IEEE Transaction on Power Systems, in print.
S. Weerasooriya and M. A. El-Sharkawi, "Dynamic Security Assessment of Power Systems Using Neural Networks," International Conference on Expert System Applications for the Electric Power Industry, December 8-10, 1993, Phoenix, Arizona.
M. A. El-Sharkawi and S. S. Huang, "Application of Genetic-Based Neural Networks to Power System Static Security Assessment," International Conference on Intelligent System Application to Power Systems, September 5-9, 1994, Montpellier, France.
M. A. El-Sharkawi and S. Weerasooriya, "Role of Neural Network in Dynamic Security Assessment: A Utility Study," International Conference on Artificial Neural Networks (ICANN), October 9-13, 1995, Invited Paper, Paris, France.
M. A. El-Sharkawi and S. .J. Huang, "Development of Genetic Algorithm Embedded Kohonen Neural Network for Dynamic Security Assessment," International Conference on Intelligent System Application to Power Systems, Jan. 28-Feb. 2, 1996, Orlando, Florida.
M. B. Zayan, M. A. El-Sharkawi and N. R. Prasad, "Comparative Study of Feature Extraction Techniques for Neural Network Classifier," International Conference on Intelligent System Application to Power Systems, Jan. 28-Feb. 2, 1996, Orlando, Florida.
Yakout Mansour, Ebrahim Vaahedi and M. A. El-Sharkawi, "Large Scale Dynamic Security Screening and Ranking Using Neural Network," V Symposium of Specialists in Electric Operational and Expansion Planning, May 19-24, 1996, Recife, Brazil.
http://cialab.ee.washington.edu/home.html-ssi
Computational Intelligent Applications Laboratory
Principal Investigator: M. A. El-Sharkawi
Sponsors: The Boeing Airplane Co., Washington Technology Center
Micro Encoder Inc.
Abstract: Controlling the rotor position is no longer the only goal in modern High Performance Drives (HPD) such as robotics, guided manipulation and supervised actuation. HPD systems are quite different from the positional control applications where only the final value of the rotor position and/or speed is controlled with no or minimal control on the traveling time or overshoots. It is essential that the rotor of the HPD system follows a preselected track at all time. A multi-robot system performing a complementing function must have the end effectors move about the space of operation according to a pre-selected time tagged trajectory. To achieve this, every motor in the robot arm must follow a specific track so that the aggregated motion of all motors keeps the end effector alongside its trajectory at all time. This must be achieved even when the system loads, inertia and parameters are varying. Any HPD system must have three basic components: an electric motor; a broad performance high speed solid-state switching converter and an advanced controller. The control strategy must be adaptive, robust, accurate, and simple to implement. Unlike most of the conventional positional controllers, the controllers for high performance tracking do not employ constant parameters. The structure and/or the parameters of the controllers must be adaptively tuned to achieve two basic objectives: 1) to provide the best possible tracking performance without overstressing the hardware; and 2) to enhance the system robustness. In the applications where the parameters of the load or the drive system are changing, the robustness of the controller is a basic requirement. Fixed parameters' controllers, such as the PID can not be considered robust. Some of the adaptive control techniques such as the variable structure and the self-tuning do not employ a physical model for the system dynamics. The dynamic model is developed based on the input/output response of the system under control. These models are usually linear but updated every several sampling intervals. Although, these adaptive controllers can be effective, they are complex to develop and require elaborate hardware to implement. With the introduction of improved training algorithms and new network topologies, the Neural Networks (NN) have demonstrated its feasibility and practicality in several applications including electric drives. Artificial Neural Network using parallel and distributed processing units can achieve the functions of system modeling and control. NN has several key features that make it highly suitable for HPD applications.
References:
A. A. El-Samahy, M. A. El-Sharkawi and S. M. Sharaf, "Adaptive Multi-Layer Self-Tuning High Performance Tracking Control for DC Brushless Motor," IEEE Transaction on Energy Conversion, June 1994, pp. 311-316.
M. A. El-Sharkawi, A. A. El-Samahy and M. L. El-Sayed, "High Performance Drive of DC Brushless Motors Using Neural Network," IEEE Transaction on Energy Conversion, June 1994, pp. 317-322.
T. C. Huang and M. A. El-Sharkawi, "High Performance Speed and Position Tracking of Induction Motors Using Multi-Layer Fuzzy Control," IEEE Transaction on Energy Conversion, June 1996, pp. 353-358.
M. Akiyama, K. Kobayashi, I. Miki, and M. A. El-Sharkawi, "Auto-Tuning Method for Vector Controlled Induction Motor Drives," The Transactions of the IEE of Japan, Volume 116-D, No. 8, 1996, pp. 844-851.
M. A. El-Sharkawi, "Neural Network Application to High Performance Electric Drive Systems," Proceedings of the IEEE Industrial Electronics Conference (IECON), November 6-10, 1995, pp. 44-49. Invited Paper, Orlando, Florida.
Tony C. Huang and M. A. El-Sharkawi, "Induction Motor Efficiency Maximizer Using Multi-Layer Fuzzy Control," International Conference on Intelligent System Application to Power Systems, Jan. 28-Feb. 2, 1996, Orlando, Florida.
A. S. Kulkarni and M. A. El-Sharkawi, "Speed Estimator for Induction Motor Drives Using an Artificial Neural Network," IEEE International Electric Machines and Drives Conference (IEMDC'97), May 18-21, 1997, Milwaukee, Wisconsin.
M. A. El-Sharkawi, T. Huang and A. El-Samahi, "Intelligent Control for High Performance Drives," Invited Paper, IEEE International Electric Machines and Drives Conference (IEMDC'97), May 18-21, 1997, Milwaukee, Wisconsin.
http://cialab.ee.washington.edu/home.html-ssi
Computational Intelligent Applications Laboratory
Principal Investigator: Mohamed A. El-Sharkawi
Sponsor: The Boeing Company
Abstract: Although the discipline has been around for quite some time, interest in the application of artificial neural networks has exploded in the last decade. Within three years, five new journals dedicated only to artificial neural networks appeared. A number of international conferences have attracted thousands of participants. Japan, Europe and the United States have each launched multi-million dollar research programs into the field of artificial neural networks and their applications. The reason for the excitement is the incredible potential computational abilities of the neural net and the ability of modern technology to implement the required neural net architectures. Neural networks have found use in numerous fields,including speech recognition, stock market forecasting, mortgage brokering, and remote sensing. Since the neural net is amenable to learning inherently nonlinear and/or complex relationships from examples, a number of system problems are potentially applicable to neural net solutions. Neural networks are especially suited for several electrical system problems such as stability assessment, harmonic evaluation and detection, fault diagnosis, adaptive control, and alarm processing. The purpose of this research is to evaluate the Artificial Neural Network technology for aviation electrical systems. This includes, electric drives control, fault detection of electrical equipment, and stability assessment.
http://cialab.ee.washington.edu/home.html-ssi
Computational Intelligent Applications Laboratory
Principal Investigator: Robert J. Marks, II
Sponsor: Boeing Information and Support Services
Abstract: This project, with the direct assistance with researchers from the Boeing Company, focuses on adaptive resonance, adaptively trained neural networks, the layered perceptron, the fuzzy paradigms and evolutionary computation applied to neural networks.
Areas currently to be developed include,
*Application of neural networks in computer imaging and CAD design;
*Application of neural evolutionary computing to parameter identification in aerospace and hydraulic and braking models;
*Establishment of stability of neuro-fuzzy control;
*Application of fuzzy systems theory to neural control and pattern recognition;
*Analysis of neural performance;
*Application of category theory to neural networks.
Other development directions may, at the agreement of both parties, be introduced during the contract period. Boeing Information and Support Services personnel will assist in coordination of this development to applications of interest to Boeing Information and Support Services.
References:
P. Arabshahi, J.J. Choi, R.J. Marks II and T.P. Caudell, "Fuzzy Parameter Adaptation in Optimization: Some Neural Net Training Examples," Computational Science and Engineering, (IEEE Computer Society), vol 3, No 1, Spring 1996, pp.57-65.
R. J. Streifel, R.J. Marks II, R. Reed. J.J. Choi and M. Healy, "Dynamic Fuzzy Control of Genetic Algorithm Parameter Coding", IEEE Transactions on Systems, Man and Cybernetics (in press)
R. von Doenhoff, R.J. Streifel and R.J. Marks II, "Carbon Brake Friction Model Parameter Identification Using Genetic Algorithms", Proceedings of the 1995 Design Engineering Technical Conferences, DE-Vol.84-1, Vol.3 - Part A, American Society of Mechanical Engineers (ASME), Boston Massachusetts, September 17-20, 1995.
Russell D. Reed and Robert J. Marks II, "An Evolutionary Algorithm for Function Inversion and Boundary Marking" Proceedings of the IEEE International Conference on Evolutionary Computation, Perth, Australia, November 26-30, 1995.
Robert J. Marks II, "Neural Network Evolution: Some Comments on the Passing Scene", Proceedings of the IEEE International Conference on Neural Networks (ICNN), pp.1-6, Washington D.C., June 2-6, 1996 - plenary paper.
R.J. Streifel, R.J. Marks II, R. Reed. J.J. Choi and M. Healy, "Dynamic Fuzzy Control of Genetic Algorithm Parameter Coding", IEEE Transactions on Systems, Man and Cybernetics (in press).
http://cialab.ee.washington.edu/home.html-ssi
Computational Intelligent Applications Laboratory
Principal Investigator: Dr. Paul S. Cho, University of Washington
School of Medicine
Robert J. Marks II, University of Washington, Department of Electrical
Engineering,
Seho Oh, Neopath, Affilate Assistant Professor, Department of Electrical
Engineering,
Shinhak Lee, Department of Electrical Engineering
Abstract: Synthesis of beam profiles for a given dose prescription is a central problem in radiotherapy. Care must be taken in the beam design to expose the tumor volume at a high level, avoid significant irradiation of critical organs, and minimize exposure of all other tissue. Use of the synthesis procedure known as alternating projections onto convex sets (POCS) is shown to be a viable approach to beam design.
POCS is a powerful tool for signal and image restoration and synthesis. Convex sets of signals obeying desired constraint sets are first specified. Then, by repeated projection onto these sets, convergence is to a signal obeying all desired constraints if the constraint sets have a finite intersection. In this paper, we apply the method of POCS to conformal radiotherapy dose computation. The performance of the method is shown through three representative examples. Contrasts to the methods of objective function minimization are to be investigated.
References:
P. Cho, S. Lee, R.J. Marks II and S. Oh, "Comparison of algorithms for intensity modulated beam optimization: projections onto convex sets and simulated annealing," Proceedings of the XII International Conference on the Use of Computers in Radiation Therapy, May, 1997, Salt Lake City.
R.J. Marks II, L. Laybourn, S. Lee and S. Oh, "Fuzzy and extra-crisp alternating projection onto convex sets (POCS)," Proceedings of the International Conference on Fuzzy Systems (FUZZ-IEEE), March 20-24, 1995, pp. 427-435, Yokohama, Japan.
S. Lee, P.S. Cho, R.J. Marks II and S. Oh, "Conformal Radiotherapy Computation by the Method of Alternating Projection onto Convex Sets," Phys. Med. Biol., July 1997.
http://cialab.ee.washington.edu/home.html-ssi
Principal Investigators: Ming-Ting Sun and Jenq-Neng Hwang
Sponsor: Computer and Communications Research Lab, Industrial Technology Research Institute, Taiwan
Abstract: Current multipoint videoconferencing systems offer very limited functionality and provide no personal presence control. In order for the multipoint videoconferencing to be more natural and effective, it is very desirable to offer per sonal presence control capabilities (e.g. each conference participant can re-size and re-position the displays to the most suitable formats). The system can be further expanded to support virtual meetings where the background of the input video can be re moved and a different background can be inserted to make the conference participants feel like they are having a conference in the same room. In this project, we will look into issues related to a flexible multipoint videoconferencing system.
References:
I.M. Pao and M.T. Sun, "Modeling DCT Coefficients for Fast Video Encoding," accepted, IEEE Transactions on Circuits and Systems for Video Technology, April 1999.
J. Youn and M.T. Sun, and C.W. Lin, "Motion Vector Refinement for Transcoding," accepted, IEEE Transactions on Multimedia, March 1999.
J. Youn, M.T. Sun, "Motion Estimation for High Performance Transcoding," IEEE Transactions on Consumer Electronics, vol. 44, no. 3, pp.649-658, August 1998.
I.M. Pao and M.T. Sun, "Statistical Computation of Discrete Cosine Transform in Video Encoders", Journal of Visual Communication and Image Representation, vol. 9, no. 2, pp.163-170, June 1998.
I.M. Pao and M.T. Sun, "Approximation of Calculations for Forward Discrete Cosine Transform", IEEE Transactions on Circuits and Systems for Video Technology, vol.8, no.3, June 1998.
Principal Investigator: Eve A. Riskin
Sponsor: Sloan Foundation
Abstract: This Fellowship supports our research projects in using VQ for image processing applications such as 8-bit color palette combination, edge detection, progressive transmission, and image restoration.
Principal Investigator: Eve A. Riskin
Sponsor: Hewlett-Packard Laboratories
Abstract: We are studying problems related to browsing image databases and generating hard copies over the Internet. In particular, we are working on providing an effortless link between the enormous number of images from, for example, the World Wide Web, and rendering devices including printers, monitors, and displays. We continuing our work in designing algorithms for embedded multilevel halftoning for both grayscale and color images and will develop new algorithms for generating and printing high quality multiresolution grayscale and color halftoned browse images.
Principal Investigator: Eve A. Riskin and Richard Ladner
Sponsor: United States Army Research Office
Abstract: This proposal addresses the transmission of images and video over noisy communication channels. Methodology and algorithms are being developed for simultaneous image compression and restoration in high noise environments, such as wireless communication. These algorithms will be based on a state-of-the-art wavelets and vector quantization image coding technique.
Intelligent Systems Lab
Principal Investigator: Linda G. Shapiro and Robert M. Haralick
Sponsor: Boeing Commercial Airplane Group
Abstract: This project is the continuation of our research in automated machine vision for manufacturing, which is being supported by Boeing Commercial Airplane Group. Our goal in this work is a vision-system development environment in which a combination of interactive and automated techniques greatly reduces the amount of custom engineering presently required for each machine vision implementation project. This years work involves the following subprojects: 1) measurement of circular regions 2) recognizing objects with planar, cylindrical, and threaded surfaces, 3) determining pose from point and ellipse correspondences, and 4) evaluating competing technologies for OCR applications.
References:
M. S. Costa, and L. G. Shapiro, "Scene Analysis Using Appearance-Based Models and Relational Indexing," IEEE Symposium on Computer Vision, 1995, pp. 103-108.
M. S. Costa, and L. G. Shapiro, "Relational Indexing," Proceeding of SSPR96, 1996, pp. 130-139.
http://george.ee.washington.edu/
Intelligent Systems Lab and Mobile Robotics Lab
Principal Investigator: Linda G. Shapiro
Sponsor: National Science Foundation
Abstract: Reconstruction of 3D environments from sensed data is a computer vision problem with important application to the area of virtual reality. Low-level techniques for acquiring, registering, and fitting 3D data have been thoroughly explored in the last few years. Prior work has, for the most part, been limited to single objects and has merely produced a surface description of the object as a whole. This research addresses the task of reconstructing entire 3D environments, which requires scene segmentation and image understanding techniques. In this work, we explore a domain-model approach to this problem. This is a knowledge-driven approach that attempts to understand the physical structure of the environment and the individual objects in the environment through models that define the physical properties and constraints of a particular domain.
References:
K. Pulli, T. Duchamp, H. Hoppe, J. McDonald, L. Shapiro, and W. Stuetzle, "Robust Approximate Meshes from a Collection of Range Maps," Proceedings of the Int'l Conference on Recent Advances in 3D Imaging and Modeling, May 1997.
http://www.cs.washington.edu/homes/shapiro/
Principal Investigator: Ming-Ting Sun and Jeng-Neng Hwang
Sponsor: QUADTEK, Redmond WA
Abstract: With the advent of digital video technology, it is now feasible to compress the video to relatively low bit-rates. The compressed video can be transported over a TCP/IP network to a distant location which does not have a range-limita tion. A multipoint remote surveillance system usually requires a large number of monitors which result in large space and high cost. There is a need to reduce the number of monitors and make the system easy to use. Also, since the available bandwidth fo r transporting video signals may be quite different depending on the specific network used, there is a need to support scalability in bit-rates. These needs may be achieved by employing scalable video coding and advanced GUI (Graphical User Interface). As the digital video technology becomes more widely used, there is also a need to develop a remote multipoint surveillance system which can handle the routing of mixed analog/digital video signals. In this project, we will look into these issues.
References:
M.T. Sun, T.D. Wu, and J.N. Hwang, "Dynamic Bit-Allocation in Video Combining for Multipoint Conferencing," IEEE Transactions on Circuits and Systems Part II, pp.644-648, May 1998.
Principal Investigator: Ming-Ting Sun
Sponsor:Tektronix
Abstract:
In many digital television applications, it is often necessary to convert the data rate of the compressed digital video from one data rate to another. In this project, we will investigate efficient means of transcoding MPEG-2 compressed digital video from one data-rate to another while maintaining as much of the picture quality as possible. We will also develop new techniques and software for implementing a cost-effective digital video transcoder.
Return to Index Page. Go forward to: Advanced Power Technologies