Low-Level Vision Color Constancy and Prior Color Naming for Application-independent Automated Linear-Time Detection and Quality Assessment of Superpixels in Multi-Source Color Images

Low-Level Vision Color Constancy and Prior Color Naming for Application-independent Automated Linear-Time Detection and Quality Assessment of Superpixels in Multi-Source Color Images

Place: Large Lecture Room  

Low-level Vision Statistical Color Constancy and Prior Color Naming for Application-Independent Automated Near Real-Time Detection and Quality Assessment of Superpixels / Texels in Any Color Image: (i) Uncalibrated / Calibrated, (ii) RGB True- or False-Color / Multi-Spectral, (iii) Spaceborne / Airborne / Terrestrial.

Since his M.S. in the late '80s Dr Baraldi (http://geog.umd.edu/facultyprofile/Baraldi/Andrea, homepage siam.andreabaraldi.com) has tackled the challenges of satellite/airborne/terrestrial image "big data" interpretation into timely, comprehensive and operational information products. Focused on sensory data interpretation, cognitive science encompasses philosophy, remote sensing, computer vision, human vision, machine learning, artificial intelligence, software engineering, geographic information science and environmental science. Starting from the peculiar CVC-UAB's expertise on low-level computer vision algorithms for statistical color constancy and prior color naming, Dr Baraldi's multi-disciplinary presentation will promote the combined exploitation of these low-level vision algorithms for automated near real-time detection and quality assessment of superpixels (texels) in multi-source color images, whether: (i) uncalibrated / calibrated, (ii) RGB true- or false-color / multi-spectral, (iii) spaceborne / airborne / terrestrial. There will be room for discussions after the talk, e.g., to look for possible synergies between the presented topics and deep convolutional neural networks.