Center for Automation Research
U. of Maryland
College Park MD 20742
Ph: (301)405-1235
Fax: (301)314-9115

9171 Springhill Lane
Greenbelt MD 20770
Ph: (301)982-4609

Research Interests

Image Processing, Computer Vision, Image and Video Compression, Pattern Recognition, Image Understanding, Image Registration, Multi-sensor Fusion and Multimedia Applications.

Education

1994 -
Ph.D. (May 1997), Electrical Engineering, University of Maryland, College Park.
Advisor: Dr. Rama Chellappa.
Thesis topic: 3-D terrain reconstruction through Synthetic Aperture Radar image analysis.

1991 - 1994
M.S., Electrical Engineering, University of Maryland, College Park.
Specialization: Communications and Signal Processing.

1986 - 1991
B.Tech., Electronics and Communication Engineering, University of Calicut, India.
Honors: Ranked 2nd in a class of approximately 100 students. Awarded the National Talent Search Scholarship by the Government of India from 1984-1990.

Work Experience

July 1992 -
Graduate Research Assistant, Center for Automation Research, University of Maryland.
Conducted research on various problems arising in Synthetic Aperture Radar (SAR) image analysis with applications to the 3-D modeling of a site. Research supported by various grants from DARPA and AFOSR. Additionally,

• Provided specialized software for segmentation and target detection to Northrop Grumman Corp., SAIC and LNK Corp.
• Collaborated in the preparation of research proposals for aerial image analysis with SRI Intl., HNC Software Inc. and LNK Corp.

Aug 1991 - May 1993
Graduate Teaching Assistant, Dept. of Electrical Engineering, University of Maryland.
Conducted undergraduate Electronics laboratory courses. Duties included supervising laboratory sessions, setting exams, and assigning letter grades.

Dissertation Related Research

• 3-D Terrain Reconstruction: Currently devising techniques to register aerial images acquired from radar or optical sensors to Digital Terrain Elevation Models (DTEM). The textural information in aerial images will be combined with the location information extracted from DTEMs, to generate 3-D site models.
• Image Registration: Developed an automatic procedure for registering multiple SAR images of a site. Registered images were used to reduce false alarms in target recognition algorithms, compute heights of 3-D structures, such as buildings and vegetation canopy, and improve segmentation.
• Generation of 2-D Site Maps from Aerial Imagery: Designed a system to detect man-made structures, such as buildings, roads, and vehicles and to delineate natural features in a single SAR image. The methodology included statistical model-based segmentation, morphological processing, feature detection and high-level reasoning.
• Knowledge-based Integration of Image Understanding Tasks: Ported the 2-D site model construction algorithm onto a system which facilitates semi-autonomous task selection, initiation and control. (Work done in collaboration with Dr. C. Shekhar)
• Image Segmentation: Developed Bayesian techniques for the segmentation of SAR intensity, complex, and polarimetric data. A posteriori smoothing involved Markov modeling followed by non-linear optimization.
• Automatic Target Detection: Implemented several algorithms for the robust detection of targets and other bright scatterers in radar images, under a variety of background clutter statistical models.

Other Research Projects

• Video Compression: Designed a video coding system based on block-matching with a DCT-based compression of the error frames. Familiar with MPEG and other video compression standards.
• Image Compression: Extended knowledge of lossless and lossy still image compression techniques, including Subband and Wavelet coding. Implemented a DCT-based still image compression scheme similar to the JPEG standard.
• Vector Quantization: A good understanding of the theory and practice of vector quantization. Implemented a pruned tree-structured vector quantizer for still image compression.
• Face Recognition: Implemented a radial-basis function Neural Network operating on features derived from frontal face images for gender classification. Aware of the state-of-the-art in this field.

Relevant Courses

Advanced Digital Signal Processing        Signal Compression and Vector Quantization
Digital Image Processing                            Pattern Recognition and Neural Networks
Image Understanding                                  Seminars in Face Recognition, Video Compression

Computer Skills

• Languages: C, C++, Pascal, Fortran; Software: Matlab, Mathematica, Khoros.

Publications

• ``Site model construction for the exploitation of E-O and SAR images,'' with R. Chellappa, Q. Zheng, C. Shekhar and R. Burlina. To appear in a book on Progress in IU, edited by O. Firschein and T. Strat, Morgan Kaufmann Publishers, Inc.
• ``Knowledge-based integration of IU algorithms,'' with C. Shekhar, R. Chellappa and M. Thonnat. In Proc. of IEEE International Conference on Pattern Recognition, Vienna, Austria, Aug. 1996.
• ``Building wide area 2-D site models from single- and multi-pass single-polarization SAR data,'' with R. Chellappa and L. M. Novak. In Proc. SPIE: Algorithms for Synthetic Aperture Radar Imagery III, Orlando, FL, April 1996.
• ``Building wide area 2-D site models from high resolution fully polarimetric Synthetic Aperture Radar images,'' with R. Chellappa. In Proc. of Intl. Symposium on Computer Vision, Coral Gables, FL, November 1995.
• ``Non-Gaussian CFAR techniques for target detection in high resolution SAR images,'' with R. Chellappa. In Proc. of IEEE Intl. Conference on Image Processing, Austin, TX, November 1994.

Professional Activities

• Member of the IEEE and its Signal Processing Society.
• Reviewed articles for the IEEE Transactions on Geoscience and Remote Sensing.