INB Lunch-Seminar

Applications of LARS-EN as a data reduction technique in image analysis

Bjarne Kjær Ersbøll
Informatics and Mathematical Modelling
Technical University of Denmark

In image analysis and in many other types of analysis one has a problem of many variables (eg. features) and relatively few observations (images). Standard ways of overcoming this difficulty usually include some sort of variable reduction technique (like variable selection and/or principal component analysis) followed by a suitable analysis on the remaining varibles.

Here we will use the Least Angle Regression and Selection with Elastic Net technique. Examples include analysis of 2D-electrophoresis gels, fungus species, and moisture of sand.

Illumination redundancy in in-line industrial inspection

Jens Michael Carstensen
Videometer A/S

To make proper radiometric measurements -- including color measurements -- with vision systems, a number of critical issues have to be handled. The pixel values typically come about through a composite of many different optical effects, like diffuse and specular reflectance, topography, fluorescence, illumination geometry, spectral sensitivity etc. The systems have to deal with heterogenous materials, which means that useful assumptions about the above effects -- like smoothness -- cannot in general be made.

To deal with these issues, the vision system, including the illumination geometry, must be carefully optimized with respect to the task at hand. Measurements must be made reproducible and traceable. Furthermore, one must provide the necessary redundancy in the imaging system to enable meaningful statistical analysis of the image data.

There are two powerful means of obtaining an effective redundancy: using multiple wavelengths or spectral sensitivity curves ("multispectral vision"), and using multiple illumination geometries ("multiray vision"). While multispectral techniques mainly focus on surface chemistry and color in a general sense, multiray techniques are more oriented towards physical surface properties like shape, topography, and gloss. Multispectral vision and multiray vision can obviously be combined to further enhance redundancy.

Zeit: Freitag, den 4. Mai 2007, 12 Uhr c.t.

Ort: Institut für Neuro- und Bioinformatik
        Seminarraum (1. OG, Raum 17),
        Ratzeburger Allee 160 (Geb. 64, 1. OG)