Expert in Computer Vision, Image Analysis, Face Recognition, Appearance Modeling & Rendering, Data Acquisition
Expert ID: 730365 Cyprus
Work by Expert has resulted in one issued and two provisional United States patents, as well as a number of publications in international, refereed conferences and journals. (These publications have had a significant impact in the computer vision community, as judged by the large number of citations of this work in subsequent scientific publications.) Applications of the developed techniques include: general face recognition (for identity verification, digital information security, law enforcement, and secure access to buildings), object recognition, surveillance, image-based rendering, and human-computer interfaces.
Specifically, in this area Expert has:
1. Developed algorithms that faithfully estimate the geometry and the spatially-varying surface reflectance (BRDF) of an object from a small number of training images captured in fixed viewpoint, but under varying, unknown lighting. These novel, uncalibrated photometric stereo algorithms incorporate the Torrance & Sparrow reflectance model, as well as the integrability constraint in the recovery of surface geometry, in an iterative, frequency-domain, optimization framework.
2. Used the recovered geometry and reflectance properties to generate synthetic, appearance-based representations of faces (namely, collections of images under variable lighting and viewpoint), significantly extrapolating from the viewing conditions in the training images.
3. Incorporated these rich, generated representations in novel, face-recognition algorithms, which have been extensively tested on large image databases that allow for the systematic study of the effects of changes in lighting and viewpoint. These algorithms perform almost without error, except on the most extreme lighting directions, and significantly outperform popular recognition methods that do not use a generative model.
4. Analysed psychophysical experiments to investigate the behavioral aspects of the ability of humans to identify objects under variable lighting conditions and to compare this ability to the performance of a computational, appearance-based model.
Because of his broad knowledge of the field, Expert can consult on many aspects of computer vision and image analysis, including calibration, low-level vision, shading and shadowing analysis, illumination, shape reconstruction, reflectance (BRDF), image synthesis, face recognition, etc.
Expert has extensive experience in appearance modeling and rendering, with emphasis on: (a) capturing and transferring context-aware appearance histories of materials, (b) image-guided geometry inference for 3D model completion, and (c) the reconstruction of integrated texture maps of large-scale structures. Specifically, Expert has:
1. Worked on modeling the appearance of aging and weathered materials, introducing the concept of “context-aware textures.” This new class of textures incorporates the relationship between time-varying appearance of weathered materials to measured context parameters that depend on geometry and environmental factors. As part of this work, a new system was developed for the capture, analysis, and synthesis of context-aware textures on new objects. The proposed concept and framework allows for the synthesis of the realistic appearance of aging materials on new objects in a way that conforms to the object’s shape and modeled environmental conditions. Context-aware textures can have a wide range of applications in the entertainment industries, but also in digital restoration and the visual arts, and are capable of generating a broad spectrum of realistic textures produced by physical phenomena.
2. Developed, in work involving 3D acquisition and modeling, a new method for filling holes in scanned 3D models using 2D images of the missing data. This method expands previous shape-from-shading approaches to take advantage of relationships that can be learned from the existing scanned data of an object and aligned images. These relationships are used to fill in missing areas, i.e., infer the missing geometry, in a manner consistent with the hole-boundary data. The advantage of this method over previous hole-filling approaches is that it approximates the actual object geometry in the missing regions, rather than just plausibly fill the holes.
3. Developed methods for creating integrated texture-maps of large-scale structures in settings where the illumination cannot be controlled, and can vary spatially, temporally, and spectrally. Methods that handle this variability have been incorporated in a system that processes multiple color images into an integrated texture. This system makes use of the laser scanner return intensity and the captured geometry, together with color balancing and mapping of illumination-corrected images onto the target geometry. Target applications of techniques that create complete, textured models of large-scale physical objects captured in uncontrolled environments, include cultural heritage documentation, the creation of 3D digital libraries, and industrial design.
Expert has extensive experience in data acquisition:
1. He designed and built a complete, first-of-a-kind, computer-controlled, camera-synchronized-lighting (30fps), image-acquisition system, creating extensive, publicly available face and object image databases with variable lighting and viewpoint. Through this work, Expert has gained extensive experience in CCD cameras, frame-grabbers, pan-tilt programming, and Linux device drivers. The collected face image databases have been used widely by the computer vision community in testing novel face-recognition algorithms.
2. He worked extensively on the development of a 3D-Model Acquisition Pipeline and performed extensive 3D scanning and model acquisition (for both small and large-scale objects) using time-of-flight and triangulation laser scanners. Through this work, Expert has gained extensive, hands-on experience in (a) 3D laser scanners, e.g. the Leica (formerly Cyrax) HDS 2500 time-of-flight scanner and the ShapeGrabber triangulation scanner, (b) Camera SDK programming, (c) High Dynamic Range imaging and tone-mapping, and (d) Calibration techniques and procedures for 3D scanners, cameras (geometric and photometric), light sources (intensity, color, and direction), and turntable axis.
Expert has extensive experience in "shape-from-x" techniques, which use one of several cues, such as stereo, shading, texture, motion, etc., to recover 3D information from 2D images. Specifically, in this area:
1. He developed novel, uncalibrated, photometric stereo algorithms that incorporate the Torrance & Sparrow reflectance model, as well as the integrability constraint, to recover the geometry and the spatially-varying surface reflectance (BRDF) of an object from a small number of training images captured in fixed viewpoint, but under varying, unknown lighting.
2. In the area of Embedded Real-time Computer Vision systems, he has published several refereed papers on implementing stereo-reconstruction algorithms on hardware (FPGA) platforms.
3. He designed and performed a sensitivity analysis of the “Tomasi and Kanade” factorization-with-missing-data method for reconstructing shape from motion.
Expert has worked on the development of a vision-based, mobile robot navigation system suitable for structured and unstructured environments. The system utilizes visual tracking and employs vision-based control to guide the motion of a robot along visually-defined trajectories.
Expert has designed and performed a sensitivity analysis of the
“Tomasi and Kanade” factorization method for
reconstructing shape from motion. (Project funded by the
Jet Propulsion Lab, Pasadena, California.) Expert has streamlined, tested, and prepared the first public
distribution of brain-surface flattening code used in
functional MRI data visualization. Expert can consult on many aspects of computer vision and image analysis, including face and object recognition, 3D shape reconstruction, reflectance (BRDF), image synthesis, 2D and 3D Data Acquisition, shading and shadowing analysis, illumination, low-level vision, calibration, etc. He specializes in the development of novel algorithms for face recognition under variable imaging conditions, the estimation of object surface geometry and reflectance from digital images, 3D model acquisition, and the recovery of scene geometry from different image cues. He can also consult on appearance modeling and rendering, including: (a) modeling the appearance histories of materials, (b) geometry inference for 3D model
completion, and (c) the reconstruction of integrated texture maps.Application areas include: face recognition, surveillance, general object recognition, image-based rendering, digital restoration and the visual arts, cultural heritage documentation, 3D digital libraries, content-based image retrieval, reverse engineering, industrial inspection and design, etc.
|Year: 2003||Degree: PhD||Subject: Electrical Engineering||Institution: Yale University|
|Year: 1999||Degree: MS, MPhil||Subject: Electrical Engineering||Institution: Yale University|
|Year: 1996||Degree: MS||Subject: Electrical Engineering||Institution: Stanford University|
|Year: 1994||Degree: BA (Hons)||Subject: Electrical and Information Sciences||Institution: University of Cambridge|
|Years: 2012 to Present||Employer: Undisclosed||Title: Director||Department:||Responsibilities: He provides technology consulting services.|
|Years: 2007 to 2011||Employer: University of Cyprus||Title: Assistant Professor and Special Scientist||Department: Dept. of Electrical and Computer Engineering||Responsibilities: He taught core-curriculum and elective undergraduate courses and laboratories. He conducted research in Computer Vision, specifically, on face recognition, on estimating scene geometry, and on embedded, real-time computer vision systems, publishing several refereed papers on implementing stereo-reconstruction algorithms on hardware (FPGA) platforms.|
|Years: 2004 to 2007||Employer: Yale University||Title: Postdoctoral Associate||Department: Dept. of Computer Science||Responsibilities: He conducted research on appearance modeling and rendering, with emphasis on capturing and transferring context-aware appearance histories of materials, on image-guided geometry inference for 3D model completion, and on the reconstruction of integrated texture maps of large-scale structures.|
|Years: 1997 to 2003||Employer: Yale University||Title: Research Assistant||Department: Dept. of Electrical Engineering||Responsibilities: He developed algorithms for face recognition under variable lighting and viewpoint, as well as methods for estimating object surface geometry and reflectance from image brightness. He designed and built a complete image acquisition system, creating extensive, publicly available face and object image databases with variable lighting and viewpoint.|
|Years: 1997 to 2000||Employer: Yale University||Title: Teaching Fellow||Department: Dept. of Electrical Engineering||Responsibilities: He supervised engineering laboratories, gave lectures, conducted recitation sessions, and advised student class projects in basic and advanced undergraduate courses.|
|Associations / Societies|
| Expert belongs to ACM, IEEE, IEEE Computer Society, Siggraph, and Eurographics.
| Expert was on the Program Committee of IEEE Conf. on Computer Vision and Pattern Recognition, IEEE Int. Conf. on Computer Vision, European Conference on Computer Vision, International Conference on Pattern Recognition, IEEE International Conference on Multimedia & Expo. He is also a reviewer for numerous journals and conferences, including IEEE TPAMI, IEEE TVCG, IJCV, JOSA, IMAVIS, ACM Computing Surveys, IEEE Trans. on Image Proc., Journal of Visual Comm. and Image Representation, CVIU, CVPR, ECCV, Eurographics, IEEE Int. Conf. on Automatic Face and Gesture Recognition.
|Awards / Recognition|
|Yale University Fellowship for graduate studies in Electrical Engineering, 1996-1997.
CASP Scholarship, Cyprus-America Scholarship Program, for graduate studies in the United States, 1994-1995.
The Institution Prize, Council of The Institution of Electrical Engineers (IEE), England, for distinction shown in the final-year project for the B.A. (Honors) degree in Electrical and Information Sciences, University of Cambridge, June 1994.
The Foundation Scholarship, Jesus College, University of Cambridge, for exceptional academic performance, June 1993.
Cambridge Commonwealth Trust Scholarship for undergraduate studies at the University of Cambridge, 1991-1994.
|Publications and Patents Summary|
| Expert has published 4 journal papers, 1 book chapter and 11 peer-reviewed conference papers, with over 2500 citations in subsequent scientific publications (Google Scholar). He has also one issued patent on Illumination-based Image Synthesis, and two provisional patents on Constructing 3D Models of Human Faces and Objects from Images with Varying Light Source Positions.
Selected Publications and Publishers:
ACM Trans. on Graphics, ACM Trans. on Applied Perception, IEEE Trans. on Pattern Analysis and Machine Intelligence, IEEE Trans. on Computers, Int. Conf. on Computer Vision, IEEE Conf. on Automatic Face and Gesture Recognition, IEEE Conf. on Computer Vision and Pattern Recognition, Int. Symp. on 3D Data Processing, Visualization and Transmission.
|Training / Seminars|
|Expert has taught core-curriculum and elective undergraduate courses and laboratories at various universities. Courses include Computer Vision, Digital Systems, Computer Organization and Microprocessors, Circuits and Electronics Laboratory, Random Signals and Systems, etc.|
|As a member of IEEE, ACM, IEEE Computer Society, Siggraph, and Eurographics, he follows developments in his areas of expertise closely and can consult on the potential of upcoming or nascent technologies.|
|Other Relevant Experience|
|COMPUTING SKILLS: Programming languages: C/C++ (incl. OpenGL), MIPS Assembly, VHDL, Pascal, Basic, HTML; Software tools: Matlab (extensive experience), Mathematica, Simulink, LaTeX, Office (PowerPoint, Word, Excel), Adobe Tools (Photoshop, Illustrator, Acrobat), Altera Quartus II, Magic VLSI layout tool, SPIM MIPS Simulator, Macromedia DreamWeaver, WinEdt, SmartDraw; Operating Systems: GNU/Linux (incl. system administration, configuration, and installation), UNIX (SunOS, IRIX, etc.), MS Windows (XP, Vista, Win 7), Cygwin, OS X, iOS, MS-DOS.
HARDWARE SKILLS: 3D laser scanners (Leica—formerly Cyrax—HDS 2500 time-of-flight scanner and ShapeGrabber triangulation scanner), CCD cameras, Camera SDK programming (Olympus SDK), High-Dynamic-Range imaging and tone-mapping (incl. HDRShop), frame-grabbers (e.g. Data Translation DT3155), pan-tilt programming, Linux device drivers, spectrophotometers (XRiteColor DTP22/Digital SwatchBook), FPGA programming. Designed and built a computer-controlled, camera-synchronized-lighting (30fps), image-acquisition system. Developed a 3D-model acquisition pipeline. Calibration for 3D scanners, cameras (geometric and photometric), light sources (intensity, color, and direction), and turntable axis.
MATHEMATICAL SKILLS: Signal and image processing, pattern recognition, linear algebra, control theory, differential geometry, probability, statistics, statistical inference, stochastic processes, optimization, applied numerical methods, digital filtering, adaptive signal processing, neural networks, vector quantization, Fourier transform, wavelets.
|Greek||He is a native speaker.|
Fields of Expertise
computer vision, machine vision, image analysis, image processing, pattern recognition, facial recognition, biometric identification, computational geometry, visible light reflectance measurement, realistic image synthesis, visual psychophysics, image rendering, computer graphics, three-dimensional computer graphics, surface texture, three-dimensional model, data acquisition, data acquisition system, imaging, 3D imaging, laser scanning system, CCD camera, digital camera, real-time imaging, imaging system, lighting system design, light control, strobe lighting system, optical measurement, calibration, three-dimensional machine vision, computer reconstruction, depth perception, binocular vision, stereo photography, embedded processing, hardware description language, real-time system, robot vision, self-navigation system, machine learning, eigenfaces, vision science, principal components analysis, Mahalonobis distance, discrete cosine transform, motion detection, JPEG standard, camera lens, dithering, calibration software, computer imaging software, aliasing, binary thresholding, optical scatter, illuminance, adaptive image filtration, image filtration, iterative method, regression analysis, video frame grabber, image processing software, image calibration, sensitivity analysis, spectrophotometer, measurement error analysis, image edge detection, image segmentation, image restoration, image enhancement, optical transfer function, Fourier analysis, photometry, color measurement, graphic object, optimization, image compression, field-programmable gate array, reflectivity, color matching, illumination measurement, artificial neural network, VHSIC Hardware Description Language, robotics, numerical optimization, numerical analysis, multivariate analysis, least squares method, graphics software, factor analysis, digital signal processing, data analysis, cluster analysis