Boaz J. Super, Ph.D.

I am now at Motorola Labs.


I am not accepting any new students.

I can still be contacted at the email address below.

Computer Science Department

University of Illinois at Chicago
Rm. 1136 SEO
851 S. Morgan St.
Chicago, IL 60607

(312) 413-8719

super 'at' cs 'dot' uic 'dot' edu

New: We've just received NIH funding to develop new image analysis methods for biomaterials research using synchrotron-based microtomography.

New: "Classification of Contour Shapes Using Class Segment Sets," by Kang B. Sun and Boaz J. Super, in CVPR 2005. Preprint.

Other Affiliations:

Adjunct Assistant Professor, Department of Ophthalmology and Visual Sciences
Assistant Professor, Department of Bioengineering
Member, UIC Vision Sciences Research Group

Editorial:

I am an Associate Editor of Pattern Recognition.
I am on the Program Comittee of the First International Conference on Complex Medical Engineering.

Teaching:

In Spring 2005 I am teaching Computer Vision I (CS 415) and C/C++ and Matlab Programming for Engineers (CS 109).

Research Interests:

Computer vision, biological vision, image processing, pattern recognition, bioengineering


Research overviews by subject with selected papers

     (Click here for a more comprehensive list of papers arranged chronologically.)

Jump to:
  1. object recognition & shape matching
  2. perceptual organization
  3. shape from texture
  4. fractals
  5. microtomography
  6. stereo
  7. color

Shape Matching and Object Recognition:

Part of object recognition is the ability to retrieve from a large database of shapes the ones most similar to the unknown object.

Our RACER system is designed from the ground up to be simultaneously accurate and fast.  Here is a snapshot of its performance on the widely-used 1400-shape MPEG-7 test database.
 

Retrieval accuracy (bullseye score) 84.05%
Classification accuracy 97.4%
Time required to match two shapes (Matlab, 1.8 Ghz Pentium PC) 3.7 milliseconds

The most recent published results are in this paper:

Boaz J. Super, "Retrieval from Shape Databases Using Chance Probability Functions and Fixed Correspondence," International Journal of Pattern Recognition and Artificial Intelligence Vol. 20, No. 8, 1117-1138, 2006. Preprint

Some of the other papers on this research:

Boaz J. Super, "Knowledge-based Part Correspondence," Pattern Recognition, 2007 (in press). Preprint

Boaz J. Super (2004), "Learning Chance Probability Functions for Shape Retrieval or Classification," IEEE Workshop on Learning in Computer Vision and Pattern Recognition (at CVPR), Washington, D.C. Download pdf

Boaz J. Super (2003), "Improving Object Recognition Accuracy and Speed through Non-Uniform Sampling," SPIE Conference on Intelligent Robots and Computer Vision XXI: Algorithms, Techniques, and Active Vision, Providence, RI, pp. 228-239. Download pdf

Boaz J. Super (2004), "Fast Correspondence-based System for Shape Retrieval," Pattern Recognition Letters 25(2) 217-225. Download pdf

Boaz J. Super (2003), "Fast Correspondence-based System for 2-D Shape Classification," The 2003 International Conference on Imaging Science, Systems, and Technology, Las Vegas, NV, June 23-26, 2003.

Boaz J. Super (2003), "Generalization Accuracy of a Fast 2-D Shape Retrieval Method," The 2003 International Conference on Imaging Science, Systems, and Technology, Las Vegas, NV, June 23-26, 2003.

Older papers on shape retrieval and object recognition using a part-indexing approach instead of RACER's holistic approach:
Boaz J. Super (2002), "Fast Retrieval of Isolated Visual Shapes," Computer Vision and Image Understanding, 85(1) 1-21. Download pdf

Boaz J. Super and Hao Lu (2003), "Evaluation of a Hypothesizer for Silhouette-Based 3-D Object Recognition," Pattern Recognition 36(1) 69-78.

Boaz J. Super (2000), "Visual Shape Retrieval Using Multiscale Term Distributions," in Proc. SPIE Conf. on Storage and Retrieval for Media Databases 2000, SPIE Vol. 3972, pp. 222-233.

Xiaonong Li and Boaz J. Super (1999), "Fast Shape Retrieval Using Term Frequency Vectors," in Proc. IEEE Workshop on Content-based Access of Image and Video Libraries, pp. 18-22.


Psychophysics Research: Perceptual Organization and Statistics of Natural Scenes

Perceptual organization refers to the human visual system's remarkable ability to find structure and organization in the patterns of light that fall in the retina.  It is an essential intermediate processing stage on the way to high-level vision tasks such as recognition.

The Psychological Review paper reports experiments we performed to measure the relative strengths of grouping by similarity of position, orientation, scale, and shape.

The Vision Research paper reports on contour grouping experiments.  The data support the explanation that contour grouping performance can be explained by a model based on the natural statistics of edge co-occurrences.

Wilson S. Geisler and Boaz J. Super (2000), "Perceptual Organization of Two-Dimensional Patterns," Psychological Review 107(4), 677-708.  Download pdf

W. S. Geisler, J. S. Perry, B. J. Super, and D. P. Gallogly (2001), "Edge Co-Occurrence in Natural Images Predicts Contour Grouping Performance," Vision Research  41(6), 711-724.  Download pdf

W. S. Geisler, J. S. Perry, B. J. Super, and D. P. Gallogly, (2000) "Natural Image Statistics Predict Contour Detection Performance," Annual meeting of the Assoc. for Research in Vision and Ophthalmology (ARVO), Ft. Lauderdale, Florida, April 30 - May 5, 2000.  Investigative Ophthalmology and Visual Science 41(4) S316.

Boaz J. Super and Wilson S. Geisler (1996), "Interaction of Similarity Dimensions in Perceptual Organization: Psychophysics and a Computational Model," ARVO, Ft. Lauderdale, Florida, April 21-26, 1996.  Investigative Ophthalmology and Visual Science 37(3) S954.

Wilson S. Geisler and Boaz J. Super (1996), "Theory of Perceptual Organization in Two-Dimensional Images," ARVO, Ft. Lauderdale, Florida, April 21-26, 1996. Investigative Ophthalmology and Visual Science 37(3) S954.

 

Shape from texture:

In some cases the 3-D shape of a physical surface can be recovered by analyzing the texture in an image of that surface.

The papers below modeled the relationship between local spatial frequency structure on the surface and the corresponding structure in images of the surface, and used AM-FM demodulation and wavelets to solve the inverse problem.

Some of these papers are for curved surfaces; others are for recovering the orientation of a planar textured surface in space.

Boaz J. Super and Alan C. Bovik (1995a), "Shape from Texture Using Local Spectral Moments," IEEE Transactions on Pattern Analysis and Machine Intelligence 17(4), 333-343.  Download pdf

Boaz J. Super and Alan C. Bovik (1995b), "Planar Surface Orientation from Texture Spatial Frequencies," Pattern Recognition 28(5) 728-743. Download pdf

Boaz J. Super (1992b), "Filters for Directly Detecting Surface Orientation in an Image," in Proc. SPIE Conf. on Visual Communications and Image Processing, SPIE Proc. Vol. 1818, pp. 144-155.

Boaz J. Super and Alan C. Bovik (1992a), "Shape-from-Texture by Wavelet-based Measurement of Local Spectral Moments," in Proc. IEEE Conf. on Computer Vision and Pattern Recognition,pp. 296-301.

Boaz J. Super and Alan C. Bovik (1991b), "Three-Dimensional Orientation from Texture Using Gabor Wavelets," in Proc. SPIE Conf. on Visual Communications and Image Processing, SPIE Proc. Vol. 1606, pp.  574-586.

Boaz J. Super, Alan C. Bovik, Wilson S. Geisler (1993), "A Model of Shape from Texture Using Second-Order Moments of Local Spatial-Frequency Spectra," Annual meeting of the Assoc. for Research in Vision and Ophthalmology (ARVO), Sarasota, Florida, May 2-7, 1993.  Investigative Ophthalmology and Visual Science 34(4) 1130.

Boaz J. Super and Alan C. Bovik (1991), "Local Image Analysis by Gabor Wavelets," IEEE Signal Processing Society 7th Workshop on Multidimensional Signal Processing, Lake Placid, NY, p. 9.13.

 

Miscellaneous: Fractals, Microtomography, Stereo, Color,

Some phenomena in nature can be described by fractals, mathematical objects with fractional dimension.  I wrote these papers when I was starting out as a graduate student.  Now I am doing fractal image analysis again as part of a collaborative project to study fracture toughness of different materials (paper in submission).
Boaz J. Super and Alan C. Bovik (1991), "Localized Measurement of Image Fractal Dimension Using Gabor Filters," Journal of Visual Communication and Image Representation 2(2) 114-128.

Boaz J. Super and Alan C. Bovik (1990), "Optimally Localized Estimation of the Fractal Dimension," in Proc. SPIE Conf. on Curves and Surfaces in Computer Vision and Graphics, SPIE Proc. Vol. 1251, pp. 357-368.
 

High-resolution 3-D imaging using synchrotron-based x-ray tomography is an important new tool of materials science.   We are currently using microtomography to study how dental materials fail.
James L. Drummond, Francesco DeCarlo, and Boaz J. Super, "3-Dimensional Tomography of Composite Fracture Surfaces," Journal of Biomedical Materials Research, Part B, Applied Biomaterials, in press.

James L. Drummond, Michael Thompson, and Boaz J. Super, "Fracture surface examination of dental ceramics using fractal analysis," Dental Materials, in press.

Using two cameras (binocular vision, or stereo) allows reconstruction of surface geometry.
Boaz J. Super and William N. Klarquist (1997), "Patch-based Stereo in a General Binocular Viewing Geometry," IEEE Transactions on Pattern Analysis and Machine Intelligence 19(3) 247-253. On-line version

Boaz J. Super and William N. Klarquist (1995), "Patch Matching and Stereopsis in a General Stereo Viewing Geometry," in Proc. International Conference on Digital Signal Processing, pp. 500-505.

Boaz J. Super (1995), "Inter-Image Metric Deformation and Sampling in Human Binocular Vision," ARVO, Ft. Lauderdale, Fort Lauderdale, Florida, May 14-19, 1995.  Investigative Ophthalmology and Visual Science36(4) S366 (abstract).

Tieh-Yuh Chen, Alan C. Bovik, Boaz J. Super (1994), "Multiscale Stereopsis Via Gabor Filter Phase Response," Proc. IEEE International Conf. on Systems, Man, and Cybernetics, pp. 55-60.

This paper evaluated the accuracy of skin detection under different color models.
Benjamin D. Zarit, Boaz J. Super, and Francis K. H. Quek (1999), "Comparison of Five Color Models in Skin Pixel Classification" Proc. International Workshop on Recognition, Analysis, and Tracking of Faces and Gestures in Real-Time Systems, pp. 58-63.
 



Useful links

Boaz's LinkedIn profile

Computer Vision Home Page

CVonline - Compendium of Computer Vision

Computer Vision Conferences

Keith Price's Annotated Computer Vision Bibliography

Biological Vision Science

ViperLib - vision illustrations and images

My Ph.D. institution: University of Texas at Austin

My Undergraduate institution: Princeton University

 
 

Copyright (c) 2002-2004 by Boaz Super.  All rights reserved.
All information on this website is protected by copyright.
No information from this website may be copied or distributed without written permission.