DAVID CASASENT: Research Description

Keywords: Computer vision, image processing, pattern recognition, neural networks, digital signal processing, detection, classification, biometrics, face and fingerprint recognition, distortion-invariant object recognition, data fusion, feature extraction, feature selection, automatic target recognition, product inspection, robotics, optical data processing.

Industrial/Commercial Relevance: Our algorithms, software and hardware have commercial use in computer vision, robotics, product inspection, target recognition, etc.

My expertise includes most areas of digital image processing and computer vision. I perform detection (location of objects or regions of interest (ROIs) in a scene; the objects can be at any orientation, contrast or class), image enhancement, feature extraction (of parameters that describe the contents of an input ROI), classification of each input object and estimation of its pose.

I apply new techniques to a diverse set of applications: robotics, active vision, product inspection, locating defects and product inspection for USDA uses, morphology, mine detection, target recognition. I process data from a wide range of sensors: visible, infrared, synthetic aperture radar, sonar, radar, ladar, acoustic backscatter, etc.

One of my main current interests is recognition of multiple objects and clutter in the face of 2-D and 3-D geometrical distortions using multiple advanced sensors. Other main present research interests concern hyperspectral product inspection, biometric recognition and verification, hierarchical classifiers, electro-optical, infrared, and synthetic aperture radar processing for object classification.