David Casasen

Prof. David Casasent

Headshot of David Casasent
Laboratory for Optical Data Processing
Department of Electrical and Computer Engineering
Carnegie Mellon University
Pittsburgh, PA 15213-3890
Telephone 412.268.2464; Fax 412.268.6345
casasent AT ece DOT cmu DOT edu

[ECE Faculty Page]
Office:
Hamerschlag Hall, B202
Administrative Assistant:
Marilyn L. Patete marilynp AT andrew DOT cmu DOT edu
Hamerschlag Hall B211, (Voice) 412-268-8162, (Fax) 412-268-6345
Research Interests:
· Image processing, distortion-invariant detection and pattern recognition
· Target recognition, robotic vision, product inspection, biometric recognition
· Neural networks, support vector machines, morphological, and distortion-invariant filters
· Nonlinear algorithm fusion to improve PD and PFA
· Synthetic aperture radar, IR and EO object detection and recognition
· Hyperspectral product inspection


Current Projects


Product Inspection - We are developing non obtrusive methods to locate anomalies in various products (e.g. locating tumors in chicken carcasses), using hyperspectral data.

Automatic Target Recognition using Distortion Invariant Filters – We locate objects or regions of interest (ROIs) in a scene; the objects can be at any orientation, contrast or class, and in IR, EO or SAR imagery.


Face Recognition using Correlation Filters – We are developing face recognition systems that are invariant to illumination, pose and e x p r e s s i o n variations using various distortion invariant filters.


Fingerprint Recognition-We have developed elastic distortion-invariant correlation filters for live-scan fingerprint matching.



Biometric Fusion – We are developing novel fusion methods which combine face and fingerprint biometric results using fingerprint data quality information, which outperforms standard multimodal results and unimodal systems.


Support Vector Machines and Neural Networks- We use support vector machines and neural networks for a wide variety of applications ranging from automatic target recognition( ATR) to biometric identification.


Not



Select Publications:
David Casasent and Yu-Chiang Wang, “A Hierarchial Classifier Using New Support Vector Machines for Automatic target Recognition”” , Neural Networks Special Issue, July 2005,pp 541-548
Rohit Patnaik and David Casasent, “ MINACE filter classification algorithms for ATR using MSTAR data”, Proc. SPIE, vol. 5807, pg 100-111,April 2005
Rohit Patnaik and David Casasent, “Illumination invariant face recognition and impostor rejection using different MINACE filter algorithms””, Proc of SPIE, vol. 5816, April 2005,pp 94-105
Chao Yuan and David Casasent, “ Face Recognition and Verification with Pose and Illumination Variations and Impostor Rejection”, keynote address, , Proc of SPIE, vol. 5779, April 2005,pg 247-255
Craig Watson and David Casasent, “Recognition of live-scan fingerprints with elastic distortions using correlation filters”, Optical Engineering, October 2004, pp 2274-2282
David Casasent and Xue-Wen Chen,“ Waveband selection for hyperspectral data: optimal feature selection”, Proc of SPIE 5106, October 2003, pp 259-270

Songyot Nakariyakul and David Casasent, “ Hyperspectral Feature Selection and Fusion for Detection of Chicken Skin Tumors” , Proc of SPIE 5271, October 2003, pp 128-139

Songyot Nakariyakul and David Casasent,“ Hyperspectral Ratio Feature Selection: Agricultural Product Inspection Example”, Proc of SPIE 5587, October 2004, pp 133-143

Selected algorithm code:
Adaptive branch and bound feature selection algorithm (created by Songyyot Nakariyakul): ABB codes.zip

Research Description
Vita: Post Script format
Resume: pdf format Postscript
Current Teaching
· 18-793 Optical Image and Radar Processing (Spring)
· 18-551 Digital Communications and Signal Processing Systems Design (Fall)

Current graduate students:
Songyot Nakariyakul, snakariy AT andrew DOT cmu DOT edu
Rohit Patnaik, rpatnaik AT cmu DOT edu
Yu-Chiang Wang, ycwang AT cmu DOT edu
Chao Yuan
Avinash Nehemiah, avinash AT cmu DOT edu