This research will prove the feasibility of a new family of smart vision chips and ancillary software, capable to adaptively select image processing parameters and regions of interest in the field of view. The proposed solution allows imaging under photon-starved conditions (low illumination, or rapidly moving objects), as well as separate processing of "bright spots". Unsupervised computer vision algorithms and software will be developed to control the real-time imaging process, using data received from the smart vision chip and/or from other modalities. The Image Algebra C++ Class Library, iac++, developed at University of Florlda under the sponsorship of DoD will be used. The silicon retina-like, smart vision chip includes on-chip signal detection and processing, while drastically reducing the electronic circuitry complexity. Superior optical resolution is possible even when manufactured with lower complexity IC technology, which may lead to photodetectors with large, wafer-scale, sensitive area. To prove the feasibility of this type of sensor, we will build an electro-optical joystick, based on the proposed technology, to be used for tests to define the design parameters for the Phase II imaging devices. The joystick has early commercialization potential in application areas such as EMI/EMP-immune vehicle position control and target acquisition and firing. Space-based use requiring miniaturization, low parts count, and EMI/EMP immunity. The joystick should find many applications in vehicle control of all kinds, such as aircraft, tanks, or even eye-movement control of vehicles or computer equipment. The vision chips to be developed in Phase II will become affordable alternatives to CCD or CID systems, especially where larger sensitive areas are desirable (target acquisition, biomedical, machine vision).
Keywords: Vision Chips, Image Algebra, Optical Joystick, Emi / Emp Hardening
