A neuro-fuzzy real-time image processing system

This paper presents a real-time image processing system based on the PAPRICA-3 linear array processor. The system comprises a PCI board, an object oriented software library and a development environment running under Windows 95/NT or Linux on a PC. The PCI board is equipped with up to 4 PAPRICA-3 chips (128 processing elements). The board features local image and program memory, the ability to grab images from several video sources through an interchangeable daughter-board and high speed hardware processing of fuzzy image matching instructions. The array processor runs from a 100 MHz-clock and is able of nearly one operation per clock cycle affecting a complete image line. The application that drove the design was the real-time recognition of the amount handwritten on banking checks, but the system can perform efficiently any kind of low-level image processing and pattern recognition