CVIP Lab Report 2003

CVIP Laboratory Phone: (502) 852-7510,

(502) 852-2789, (502) 852-6130 Fax: (502) 852-1580

cvip@cairo.spd.louisville.edu http://www.cvip.louisville.edu

About CVIP… Mission

he Commuter Vision and Image Processing (CVIP) Laboratory was established in 1994 at the University of Louisville and is committed to excellence in research and teaching of computer vision and its application.

The CVIP has two broad focus areas: computer vision and medical imaging. The laboratory hosts unique and modern hardware for imaging, computing and visualization. This hardware includes two supper computers from SGI (an 40-CPU ONYX2-R1200 and 24-CPU ONYX-10000), an ImmersaDesk from Fake-Space/Pyramids Systems, a 3-D scanner from CyberWare, a robotic arm M6-i from Fanuk, over 20 high-end graphics of workstations, and various imaging hardware. The laboratory is housed in a modern state-of-the-art research building and is linked, via a high-speed network, to the university’s medical center. Among the active research projects at the laboratory are the following:

1. Trinocular active vision, which aims at creating accurate 3-D model of indoor environments. This research is leading to creation of the UofL CardEye active vision system, which is our research platform in advanced manufacturing and robotics.

2. Multimodality image fusion, which aims at creating robust target models using multisensory information.

3. Building a functional model of human brain based on integration of structural information (from CT and MRI) and functional information

T

CCoommppuutteerr VViissiioonn

Providing a better understanding of the Human and Computer

Vision Systems.

(from EEG signals and functional-MRI scans). The functional brain model is our platform for our brain research in learning, aging, and disfunctions.

4. Image-guided minimally invasive endoscopic surgery which aims at creating a system to assist the surgeons locate and visualize, in real-time, the endoscope’s tip and field of view during surgery.

5. Large-scale visualization for modeling and simulations of physical systems, and applications in visual reality.

6. Building a computer vision-based system for reconstruction of human jaw using intra oral video images. This research will create the UofL Dental Station, which will have various capabilities for dental research and practice.

7. Vision-based system for autonomous refueling 8. Vision-based system for autonomous vehicle navigation 9. Theoretical work in image modeling, segmentation, registration and

pattern recognition. About The University of Louisville

he University of Louisville is a state supported urban university located in Kentucky’s largest metropolitan area. It was a municipally supported public institution for many decades prior to joining the university system

in 1970. The university has three campuses. The 177-acre Belknap Campus is three miles from downtown Louisville and houses seven of the universities eleven colleges and schools. The Health Science Center is situated in downtown Louisville’s medical complex and houses the university’s health related programs and the University of Louisville Hospital. On 243-acre Shelby Campus located in eastern Jefferson County are the National Crime Prevention Institute and the University Center for Continuing and Professional Education. In recent years, the university has also offered expanded campus courses at Fort Knox, Kentucky.

About The J. B. Speed School of Engineering

ounded in 1924, the University of Louisville Speed Scientific School (recently re-named The J.B. Speed School of Engineering) is the University’s college of engineering and applied sciences. Endowments

from the James Breckinridge Speed Foundation, named after an influential Louisville industrialist (1844-1912), started and have continually supported the Speed School. The School consists of Chemical Engineering, Civil & Environmental Engineering, Computer Engineering and Computer Science, Electrical and Computer Engineering, Industrial Engineering and Mechanical Engineering programs.

T

F

CVIP Laboratory/Lutz Hall, University of Louisville

1. Scene Description from Sequence of Images (sponsored by US Army)

his proposal, funded by the Mounted Battle Manuveurs Laboratory, addresses mainly a problem of dynamic scene reconstruction. In particular, we address the following problem: can an active vision

system, such as the CardEye, be used to faithfully reconstruct a 3-D description of a dynamic in-door scene at a near real-time frequency? A solution to this problem will open the door for a multitude of practical applications along the lines of what have been enumerated above. Our

T

I. Grants Abstracts

approach to solve this important and challenging problem lies in the creation of a “virtual 3-D camera” that can take “3-D reconstruction shots” of the scene at “real-time” shutter rate! The envisioned “3-D camera” is an active vision system, with proper components, mounted on a flexible mechanism, such as three-segment robotic arm. This arrangement is connected to a fast computer and visualization hardware to allow the acquisition and display of the “3-D shots.” To enable full scene construction (up to 360˚) we propose a novel design, that provides the necessary degrees of freedom in kinematics and the additional sensors to enable reliable data acquisition.

2. Novel Approaches in Perception for Autonomous Mobility (sponsored by US Army)

obotics, in general, has its impact on the society in all venues, industrial, defense, medical and the environment. The current state-of-the-art in smart robotics is quite impressive in terms of platforms and applications;

however, there exists large voids in sensors and algorithms. While modern AI robots may not be expensive, they are problem dependent and modularity is still a far away goal. The infrastructure that produces AI robots is quite involved and expensive, even though the robot itself may not be expensive at all. Such infrastructure exists only in a few major universities in the US. A major aim of this research is to lay down the foundations for smart robotics R&D at UofL. Immediate goals of this project are the following: 1) Creating a focus group of researchers, at UofL, who are involved in vision, ethology, AI, and human perception; 2) Invigorating the current research activities, at UofL, on sensor/course-of-action planning, sensor integration, and data fusion in active vision systems; 3) Integrating current resources at UofL in simulation, active vision, and computing, in order to create a comprehensive environment for research, teaching and training in AI robotics; 4) Creating a multidisciplinary curriculum in AI robotics at UofL at the senior/first-year graduate level of the engineering, computer science and psychology curricula; and 5) Addressing

R

specific needs of the US Army’s Maneuver Battle Laboratory at Fort Knox with respect to AI Robotics. Specific research tracks proposed to implement the above goals include the following: 1) Development of a virtual reality test bed for the design and evaluation of AI Robotics algorithms/architectures based on simulation; 2) Development of novel approaches for creating context sensitive labeled images for autonomous mobility; 3) Creating novel representations for reflexive and reactive behaviors; 4) Evaluation of various instances of perception-mediated reactive-behavior mobility solutions; and 5) Creating a research platform for AI Robotics that lends itself to studies in sensor planning, fusion and integration, in order to achieve optimal perception metrics for application specific AI Robots. The integrated approach attempts to address diverse aspects of the very complex, multi-faceted task of intelligence insertion into situation focused operative agents.

3. Automatic Lung Cancer Detection (sponsored by Jewish Hospital Foundation)

his research aims at developing a fully automatic Computer-Assisted Diagnosis (CAD) system for lung cancer screening using chest spiral CT scans. One thousand subjects are enrolled in a chest cancer screening

program in Louisville, KY, USA, which aims at quantification of the effectiveness of low dose spiral CT scans for early diagnosis of lung cancer, and evaluating its possible impact on improving the mortality rate of cancer patients. This research presents the first phase of an image analysis system for 3-D reconstruction of the lungs and trachea, detection of the lung abnormalities, identification/classification of these abnormalities with respect to specific diagnosis, and distributed visualization of the results over computer networks.

T

4. Multi-Modality Image Fusion for 3D Model (sponsored by Air Force Office of Scientific Research)

he recent advances in sensor systems and the increase availability of ancillary data enable one to extract far more detailed information than ever before. However, this availability of multisensor data requires the

development of effective data analysis techniques, which can utilize the full potential of the observed data. There is a consensus in the remote sensing community that one single modality cannot provide the required information to build an accurate 3D digital model of a terrain area. Therefore, multisensor integration and data fusion techniques have been used extensively in terrain mapping, classification, and 3D model building. In this project, we aim to combine/fuse multi-modality data provided from both spaceborne and airborne sensors, in order to build a 3D digital model for the sensed target area. The current progress in that project includes implementation of some algorithms for classification of remote sensing data. Also an algorithm for decision fusion is implemented.

Acknowledgments

The CVIP Lab Acknowledges the following funding sources

National Science Foundation

National Cancer Institute

Office of Naval Research

Air Force Office of Scientific Research

Norton Healthcare System

Jewish Hospital Foundation

The Whitaker Foundation

Silicon Graphics Incorporated

Special Recognition The CVIP Lab acknowledges the enormous support provided by The US Army Grant DABT60-02-P-0063. The Lab acknowledges with thanks the continued support Mr. James N. Cook, Program Manager, Experimentation and Analysis Lab, Ft. Knox, KY.

T

1. Video Reconstructions in Dentistry

Aly A. Farag, and Ahmed H. Eid,

Orthod Craniofacial Res 6 (Suppl. 1), 2003; 108–116.

II. Publications

Abstract The 3-D reconstruction of the human jaw has tremendous applications. It enables many orthodontics and dental imaging researches to be applied directly to a digital jaw model, not to a cast, using computer vision and medical imaging tools. This paper presents two practical techniques for 3-D modeling of the human jaw from a sequence of intra-oral images. The first technique is based on the shape from shading algorithm and the other one is based on the space carving (SC) algorithm for shape recovery. Shape from shading (SFS) technique, using perspective projection and camera calibration, extracts the 3-D information from a sequence of 2-D images of the jaw. Data fusion of range data and 3-D registration techniques develop the complete jaw model. The space carving approach is implemented on a sequence of calibrated images. On the two reconstructions, we fit a mesh model to the data, in order to create a solid 3-D model. These models lend themselves for various applications in density. We report experimental results that show successful application of both approaches to building 3-D models of the human jaw with sub-millimeter accuracy.

2. On The Performance Characterization Of Stereo And Space Carving

Ahmed H. Eid and Aly A. Farag

Proceedings of Acivs 2003 (Advanced Concepts for Intelligent Vision Systems), Ghent, Belgium, September 2-5, 2003

Abstract The performance evaluation of 3-D reconstruction techniques is not a simple problem to solve. This is not only due to the increased dimensionality of the problem but also due to the lack of standardized and widely accepted testing methodologies. This paper presents characterization to the performance of two standard 3-D reconstruction techniques; stereo and space carving. The evaluation is performed on the same data set using an image re-projection technique to reduce the dimensionality of the evaluation domain and convert the the problem into an image quality assessment problem. Also, different measuring strategies are presented and applied to the stereo and space carving reconstructions. These measures proved great consistency in measuring the performance of the techniques under test. The experimental results show the comparison between stereo and space carving in terms of 3-D reconstruction resolution and color inconsistency effects.

3. Automatic Identification Of Lung Abnormalities In Chest Spiral CT Scans

Ayman El-Baz and Aly A. Farag

Proc. of the International Conf. on Acoustics, speech, and signal processing, 2003.

Abstract This research aims at developing a fully automatic Computer- Assisted Diagnosis (CAD) system for lung cancer screening using chest spiral CT scans. One thousand subjects are enrolled in a chest cancer screening program in Louisville, KY, USA, which aims at quantification of the effectiveness of low dose spiral CT scans for early diagnosis of lung cancer, and evaluating its possible impact on improving the mortality rate of cancer patients. This paper presents an image analysis system for 3-D reconstruction of the lungs and trachea, detection of the lung abnormalities, identification/classification of these abnormalities with respect to specific diagnosis, and distributed visualization of the results over computer networks. We present two novel approaches for segmentation of the lung tissues from the surrounding structures in the chest cavity, and detection of the abnormalities in the lungs. The segmentation algorithm is hierarchical; it starts with isolating the background from the chest cavity, then isolating the lungs from the surrounding structures (e.g., ribs, liver, and other organs that may appear in chest CT scans). Abnormalities in the lungs are detected by analyzing the segmented lung tissues and extracting the isolated lumps that appear in various connected regions. 3-D reconstructions are also generated for these abnormalities, in order to be used for subsequent identification/classification steps. Results of these algorithms are shown on 50 subjects, and have been evaluated vs. the radiologists. The image analysis approach presented in this paper has provided comparable results with respect to the experts. The approach is quite fast, and lends itself to distributed visualization over computer networks.

4. Pararmeter Estimation In Gibbs-Markov Image Models

Ayman El-Baz and Aly A. Farag

Proceedings of 6th International Conference on Information Fusion, July 8-11, 2003, Cairns, Queensland, Australia.

Abstract In this paper we present two novel approaches to estimate the clique potentials in Gibbs-Markov image models. In the first approach, we estimates the clique potentials for binary Gibbs Markov random field (GMRF) by using genetic algorithms (GAs). While, in the second approach, we estimated the parameters of Gaussian Markov random field. The outline steps of the second algorithm are as follow. Given an image formed of a number of classes, an initial class density is assumed and the parameters of the densities are estimated using the EM approach. Convergence to the true distribution is tested using the Levy distance. The segmentation of classes is performed iteratively using the ICM algorithm though a genetic algorithm (GA) search approach that provides the maximum a posteriori probability of pixel classification. During the iterations, the GA approach is used to select the model order and the corresponding clique potentials of the Gibbs-Markov models used for the observed image. The approach has been tested on real and synthetic images and provided satisfactory results.

5. A New Unsupervised Approach For The Classification Of Multispectral Data

Refaat M. Mohamed and Aly A. Farag

Proceedings of 6th International Conference on Information Fusion, July 8-11, 2003, Cairns, Queensland, Australia.

Abstract In this paper, we present a new approach for unsupervised pixel-wise classification of multispectral data. In this approach, we estimate the number of classes in a data set as well as the parameters of each class. This approach doesn’t require training samples. A scene is modeled as a realization of a random field formed of a mixture distribution. We employ an EM algorithm to estimate the parameters of this mixture. The algorithm iterates to estimate the proper number of classes that compose the scene, and also the parameters of these classes. Theoretically, the algorithm can be applied for arbitrary forms of distributions for the classes in the scene, either parametric or non parametric. This paper focuses on the parametric estimation case. The approach has been applied to synthetic and real world multispectral data and provided quite encouraging results compared to other classical classification approaches.

6. Statistical-Based Approach For Extracting 3D Blood Vessels From TOF-MRA Data.

M. Sabry Hassouna, Aly A. Farag , Stephen Hushek, and Thomas Moriarty

To appear in International Conference on Medical Image Computing and Computer-Assisted Intervention (MICCAI'03), Montréal, Québec, Canada, 15-18 November 2003.

Abstract In this paper we present an automatic statistical intensity basedapproach for extracting the 3D cerebrovascular system from time-of-flight (TOF) magnetic resonance angiography (MRA) data. The voxels of the dataset are classified as either background tissues, which are modeled by a finite mixture of one Rayleigh and two normal distributions, or blood vessels, which are modeled by one normal distribution. We show that the proposed models fit the clinical data properly and result in fewer misclassified vessel voxels. We estimated the parameters of each distribution using the expectation maximization (EM) algorithm. Since the convergence of the EM is sensitive to the initial estimate of the parameters, a novel method for parameter initialization, based on histogram analysis, is provided. A new geometrical phantom motivated by a statistical analysis was designed to validate the accuracy of our method. The algorithm was also tested on 20 in-vivo datasets. The results showed that the proposed approach provides accurate segmentation, especially those blood vessels of small sizes.

MRA Segmentation Results

7. Image Segmentation Using Gibbs-Markov Models: Parameters Estimation And Applications

Ayman El-Baz and Aly A. Farag

IEEE International Conference on Image Processing, Sep. 14-17, 2003 Barcelona, Spain.

Abstract Stochastic models of images are commonly represented in terms of three random processes (random fields) defined on the region of support of the image. The observed image process G is considered as a composite of two random process: a high level process Gh , which represents the regions (or classes) that form the observed image; and a low level process Gl , which describes the statistical characteristics of each region (or class). The representation G = (Gh , Gl ) has been widely used in the image processing literature in the past two decades. In this paper, we consider the low level process Gl as mixture of normal distributions, and we use the Expectation-Maximization (EM) algorithm to estimate the mean, the variance, and proportion for each distribution. A popular model for the high level process Gh has been the Gibbs-Markov random field (GMRF) model. We introduce a novel unsupervised approach to estimate the parameters of a GMRF model. In this approach, we estimate the model parameters that maximize the posteriori probability of each pixel in a given image. The MAP estimate is obtained using a combination of genetic search and deterministic optimization using the iterated conditional mode (ICM) approach of Besag. The desired estimate of the GMRF parameters is the one corresponding to the MAP estimate. The approach has been applied on real images (Spiral CT slices) and provides satisfactory results.

8. MRA Data Segmentation Using Level Sets

Hossam Hassan and Aly A. Farag

IEEE International Conference on Image Processing, Sep. 14-17, 2003 Barcelona, Spain.

Abstract In this paper, we use a level set based segmentation algorithm to extract the vascular tree from Magnetic Resonance Angiography, “MRA”. Classification model finds an optimal partition of homogeneous classes with regular interfaces. Regions and their interfaces are represented by level set functions. The algorithm initializes level sets in each image slice using automatic seed initialization and then iteratively, each level set approaches the steady state and contains the vessel or non-vessel area. The algorithm is applied on each slice of the volume to build up the tree. The results are validated using a phantom that simulates the “MRA”. The approach is fast and accurate. Results on various cases demonstrate the accuracy of the approach.

MRA Segmentation Results

9. Two Sequential Stages Classifier for Multispectral Data

Refaat M Mohamed and Aly A Farag

Proceedings of the International Conference on Computer Vision and Pattern Recognition (CVPR) 2003 workshop on Intelligent Learning, Madison, WS, June 16-22, 2003.

Abstract In this paper, we present an approach for the classification of remote sensing multispectral data, which consists of two sequential stages. The first stage exploits the capabilities of the Support Vector Machines (SVM) approach for density estimation and uses it in a Bayes classification setup. In a typical image, the class of a pixel is highly dependent on the classes of its neighbor pixels. The second stage of our classifier applies for this dependency of the class. We incorporate this dependency using stochastic modeling of the context as a Markov Random Field (MRF). The MRF is modeled using Besag model and implemented using the Iterative Conditional Modes (ICM) algorithm. Results show that the stochastic modeling approach enhances the results and provides reasonable smoothness in the classified image.

Classification Results for Golden Bay Area

10. Classification of Multispectral Data Using Support Vector Machines Approach for Density Estimation

Refaat M Mohamed and Aly A Farag

Proceedings of the International Conference on Computer Vision and Pattern Recognition (CVPR) 2003 workshop on Intelligent Learning, Madison, WS, June 16-22, 2003.

Abstract this paper, we present an approach for the classification of remote sensing multispectral data, which exploits the capabilities of the Support Vector Machines (SVM) approach for density estimation. Extending the support vector machines to estimate multidimensional densities is explored. We use these estimates in the design and implementation of Bayes classification of multispectral Landsat data. Density estimation using SVM is compared with two traditional approaches, the Parzen window and k-NN approaches. Results on synthetic and real world remote sensing data show that the SVM estimates are more superior to the other methods in terms of accuracy, robustness and convergence speed.

Classification Results from Agricultural Area

11. Parameter Estimation for Bayesian Classification of Multispectral Data

Refaat M Mohamed and Aly A Farag

Proceedings of the International Conference on Computer Vision and Pattern Recognition (CVPR) 2003 workshop on Intelligent Learning, Madison, WS, June 16-22, 2003.

Abstract In this paper, we present two algorithms for estimating the parameters of a Bayes classifier for remote sensing multispectral data. The first algorithm uses the Support Vector Machines (SVM) as a multi dimensional density estimator. This algorithm is a supervised one in the sense that it needs in advance, the specification of the number of classes and some training samples for each class. The second algorithm employs the Expectation Maximization (EM) algorithm, in an unsupervised way, for estimating the number of classes and the parameters of each class in the data set. Performance comparison of the presented algorithms shows that the SVM- based classifier outperforms those based on Gaussian based and Parzen window algorithms. We also show that the EM based classifier provides comparable results to Gaussianbased and Parzen window-based while is an unsupervised.

12. Stochastic Models in Image Analysis: Parameter Estimations and Case Studies in Image Segmentation

Ayman El-Baz and Aly A Farag

IEEE workshop on statistical signal processing, Sept, 2003

Abstract Stochastic models of images are commonly represented in terms of three random processes (random fields) defined on the region of support of the image. The observed image process G is considered as a composite of two random process: a high level process X , which represents the regions (or classes) that form the observed image; and a low level process Y , which describes the statistical characteristics of each region (or class). The representation G = (X, Y) has been widely used in the image processing literature in the past two decades. In this paper we will show how to use Expectation Maximization (EM) algorithm to get accurate model for the low level image by using mixture of normal distribution. The main idea of the proposed algorithm is as follow: first, we will use the EM algorithm to get the most dominance mixtures in the given density (Empirical density), and then we will assume that the absolute error between the empirical density and the estimated density is another density and we will use the EM algorithm to estimates the number of mixtures in this error and the parameters for each mixtures. Then the estimated density for the absolute error is added or subtracted from the estimated density according to the sign(error). Convergence to the true distribution is tested using the Levy distance. A popular model for the high level process X has been the Gibbs-Markov random field (GMRF) model. In this paper we will use the same approach which described in [1] to estimate the parameters of GMRF. The approach has been applied on real images (Spiral CT slices) and provides satisfactory results.

13. A Unified Approach for Detection, Visualization, and Identification of Lung Abnormalities in Chest Spiral CT Scans

Ayman El-Baz and Aly A Farag

Proc. Of Computer Assisted Radiology and Surgery, London, June, 2003.

Abstract This research aims at developing a fully automatic Computer-Assisted Diagnosis (CAD) system for lung cancer screening using chest spiral CT scans. One thousand subjects are enrolled in a chest cancer screening program in Louisville, KY, USA, which aims at quantification of the effectiveness of low dose spiral CT scans for early diagnosis of lung cancer, and evaluating its possible impact on improving the mortality rate of cancer patients. This program is also a member of a national society program (ACRUN). This paper presents the first phase of an image analysis system for 3-D reconstruction of the lungs and trachea, detection of the lung abnormalities, identification/classification of these abnormalities with respect to specific diagnosis, and distributed visualization of the results over computer networks. We present two novel approaches for segmentation of the lung tissues from the surrounding structures in the chest cavity, and detection of the abnormalities in the lungs. The segmentation algorithm is hierarchical; it starts with isolating the background from the chest cavity, then isolating the lungs from the surrounding structures (e.g., ribs, liver, and other organs that may appear in chest CT scans) by using Gibbs Markov Random Field (GMRF). Abnormalities in the lungs are detected by analyzing the segmented lung tissues and extracting the isolated lumps that appear in various connected regions. In order to detect these abnormalities we developed template-matching technique based on a genetic algorithm (GA) template matching (GATM) for detecting nodules existing within the lung area. The GA was used to determine the target position in the observed image efficiently and to select an adequate template image from several reference patterns for quick template matching. In addition, a conventional template matching was employed to detect nodules existing on the lung wall area, lung wall template matching (LWTM), where semicircular models were used as reference patterns; the semicircular models were rotated according to the angle of the target point on the contour of the lung wall. After initial detecting candidates using the two template matching methods, we extracted a total of 12 feature values and used them to eliminate false-positive findings using Bayesian’s classifier. Also 3-D reconstructions were generated for these abnormalities, in order to be used for subsequent identification/classification steps. Results of these algorithms are shown on 50 subjects, and have been evaluated vs. the radiologists. The image analysis approach presented in this paper has provided comparable results.

Aly Farag was appointed as University Scholar: Dr. Farag, Provost Garrison

(on the podium) and Dr. Nancy Martin,Vice Presdient for Research.

Aly Farag, Research Louisville, won the first prize in research for his innovations in Biomedical

Engineering. Shown, Dr. Farag and this year's winners.

III. Awards

Ayman El-Baz (middle), ECE Ph.D. student and RA at the CVIP Lab, won the first place award in student competition, Research Louisville 2002, and Dr. Chenoweth, Assistant Vice President

for Research and Chairman of the ECE Department (left) and Dr. Farag (right).

Mr. Ahmed Eid is receiving the SGI Award from Dr. Darrel Chenoweth.

Mr. Sabry receiving the SGI Award from Dr. Darrel Chenoweth.

Aly A. Farag was educated at Cairo University (B.S. in ElectricalEngineering), Ohio State University (M.S. in Biomedical Engineering),University of Michigan (M.S. in Bioengineering), and PurdueUniversity (Ph.D. in Electrical Engineering). Dr. Farag joined theUniversity of Louisville in August 1990, where he is currently aProfessor of Electrical and Computer Engineering. His researchinterests are concentrated in the fields of Computer Vision andMedical Imaging. Dr. Farag is the founder and director of the

Computer Vision and Image Processing Laboratory (CVIP Lab) at the University ofLouisville, which supports a group of over 20 graduate students and postdocs. Dr.Farag's contribution has been mainly in the areas of active vision system design, volumeregistration, segmentation and visualization, where he has authored or co-authored over80 technical articles in leading journals and international meetings in the fields ofcomputer vision and medical imaging. Dr. Farag is an associate editor of IEEETransactions on Image Processing. He is a regular reviewer for a number of technicaljournals and to national agencies including the NSF and the NIH. He is a Senior Memberof the IEEE and SME, and a member of Sigma Xi and Phi Kappa Phi. Dr. Farag is recentlyawarded a "University Scholar".

Darrel L. Chenoweth is a Professor in and the Chairman of the Electrical and Computer Engineering Department, and Assistant Vice President for Research at the University of Louisville. He joined the University of Louisville in 1970 after completing his Ph.D. at Auburn University. He has been involved in image processing and pattern recognition research since 1981, which is sponsored by the Naval Air Warfare Center and the Office of Naval Research. He's a Fellow of IEE.

Chuck Sites is a University of Louisville staff member for the Electrical and Computer Engineering Department. He received a Bachelor degree in Physics from the University of Louisville in 1990. He has over fifteen years of experience in the computer and electronics industry. He manages the computer systems and networks of the Electrical and Computer Engineering Department and is the System Administrator and Technical Advisor for the CVIP Laboratory.

IIII. Staff

Ahmed Eid was educated at Mansoura University, B.Sc. in Electronics Engineering with honor, Mansoura University and M.Sc.

in Elect. Comm. Engineering from the same university. He joined Mansoura University as a teaching assistant in 1996. He is currentlya teaching assistant at University of Louisville. He has enrolled in the ECE Ph.D program at UofL since Aug. 2000. His research interests are concentrated in the field of Computer Vision.

Ayman El-Baz received the B.S. in Electrical Engineering with honors from Mansoura University, Egypt in 1997, and M.S. from

same university in 2000. He joined the Ph.D. Program in the ECE Department in Summer 2001. Mr. Elbaz is aresearch assitant at the CVIP Lab working on medical imaging analysis of lung cancer. His interests include statistical modeling, genetic algorithms.

Mohamed Sabry joined the lab as a Ph.D. student in the Fall of 2001. Currently, he is working in 3D volume cerebrovascular

segmentation from MRA modality. His research interests include visualization of large-scale medical data sets, Medical Imaging, Pattern Recognition, and Image Processing. He won in 2002, the "Excellence in Visualization Award" from Silicon Graphics.

Refaat Mohamed received his B.S. in Electrical Engineering with honors from Assiut University, Egypt in 1995, and M.S. from the same university

in 2001. He joined the Ph.D. Program in the ECE Department in Fall 2001. Mr. Mohamed is a research assistant at the CVIP Lab working on Remote Sensing data analysis. His interests include statistical learning systems and multidimensional classification algorithms.

V. Students

Alaa El-Din Aly received his B.S. in Electrical Engineering with honors from Assiut University, Egypt in 1996, and M.S. from the same university

in 2000. He joined the Ph.D. Program in the ECE Department in Summer 2002.

Hossam Hassan joined the lab as a Ph.D. student in the summer of 2002. His research interests includes level sets segmentation, image processing and

computer vision.

Hongjian Shi Holds a Ph.D. in mathematics from Univ of British Columbia, Canada, amdhas joined the lab as a Ph.D. student in the fall of

2002. His research interests include application of finite element methods for studying brain deformation.

Emir Dizdarevic is studying for his masters. His research interests include robotics and artificial intelligence

Esen Yuksel is working on her Master of Science degree. She joined the lab in the Fall of 2003. Her research interests includes medical image analysis and

visualization.

Sergiec has joined the lab as a Ph.D student in the fall of 2003. His research interests includes Medical Imaging.

Chen joined the lab as a Ph.D. student in the summer of 2003. His research interests include medical imaging.

Rachid Fahmy Holds a Ph.D. in mathematics from France and has joined the lab as a Ph.D. student in the fall of 2003. His research interests include application of finite element methods for studying brain deformation.

“Medical imaging is computational intensive, and the need for more detailed visualization continues to be a significant issue in our research . .”

Aly Farag

Professor & Director of CVIP Lab

Success Story (http:// www.sgi.com/pdfs/3349.pdf)

Computer Vision and Image Processing Laboratory (CVIP Lab) University of Louisville, Louisville, KY 40292

Phone: (502) 852-7510, 2789, 6130 Fax: (502) 852-1580

Email: cvip@cvip.uofl.edu http://www.cvip.uofl.edu