Computer Vision Research UNR George Bebis Computer Vision

Computer Vision Research @ UNR George Bebis Computer Vision Laboratory (CVL) Department of Computer Science and Engineering University of Nevada, Reno, USA 1

Computer Vision Laboratory (CVL) Funding • Founded in 1998 to conduct basic and applied research in computer vision. • Members - 2 faculty 6 Ph. D students 4 MS students Several undergraduates 2 visiting faculty Funding: $6 M Collaborators LLNL

Main Research Areas Biometrics Segmentation Tracking Segmentation 3 D reconstruction 3 D object recognition Human action/activity recognition Applications 3

Computer Vision – Sample Projects Face Recognition Better handle changes due to lighting, facial expressions, and eye-glasses, by fusing visible with thermal infrared imagery. Gender Classification Use Genetic Algorithms to select gender-specific features. Face detection in the near-IR Use near-IR for face detection.

Computer Vision – Sample Projects Peg-less System Hand-based Authentication Person authentication using hand geometry based on high order Zernike Moments. Extended for gender classification Traditional System Patent awarded in Feb 2014

Computer Vision – Sample Projects Fingerprint Recognition Small overlapping area Minutiae Point matching Missing/spurious minutiae

Computer Vision – Sample Projects Fingerprint Mosaicking Better deal with small overlapping area and missing/spurious minutiae. m. Set 1 2 i n Super template

Computer Vision – Sample Projects Where Am I? Rover Localization Estimation of rover 3 D position and orientation using visual information and Digital Elevation Maps (DEMs). Crater Detection Detect craters using visual information. Smart Monitoring of Complex Public Scenes Automatic understanding of video content for detecting activities of interest (e. g. , potential threats). Context-Based Intent Understanding Automatic behavior modeling for effective detection of intentions.

Computer Vision – Sample Projects Intelligent Vehicles Vehicle detection for driver assistance. Image Forgery Detection Determine the authenticity of an image. Object Detection and Tracking Detect and track humans and other objects of interest. Blood Vessel Segmentation Segment blood vessels in retinal images.

Detect Natural Shapes in Cluttered Backgrounds Input: oriented segments

Iterative Multi-Scale Tensor Voting (IMS-TV)

For more info … • Visit http: //www. cse. unr. edu/CVL 12