Computational Optical Sensing and Processing Laboratory
The main research challenge of our laboratory is to derive precise abstract information or decisions from large complex noisy topological data sets captured by one or multiple optical sensors. We apply special optical arrangements such as different holographic setups and fluorescent illuminations for microscopic imaging, or multi-spectral camera-based patient monitoring systems or wide-angle multi-camera systems in monitoring. The heavy computational load is handled by many-core processor arrays, such as GPUs in desktop applications or embedded low-power systems.
Publication date
1992
MTA SZTAKI, Budapest, pp. 27.
Using CNN to "see" random-dot stereograms - dual CNN models of stereo vision. (Reseach report of the Dual and Neural Computing Systems Laboratory DNS-7-1992.)
The dual CNN analog software - a programmable analog, nonlinear, dynamic, 3D computing array plus logic to form a "multi-screen theater" on silicon. (Reseach report of the Dual and Neural Computing Systems Laboratory DNS-2-1992.)