Moving-images time gradient implementation using RAM-based FPGA

ABSTRACT Time gradient can be used to extract information from motion. It has been already done, but higher performancesare reached if it's computed on real time (at video rate). We propose an architecture to evaluate it from consecutive imagesin a video signal. Behind a delaying structure, a circuit operates over the neighbourhood of pixels around each one: not onlythe adjacent pixels in space but also in time (in several previous images). Gradient can be extracted from all of them byconvolution, and other non-linear algorithms can also be applied. Known 3x3 masks to operate over static images aregeneralized by 3D masks (3x3x3) to operate over dynamic images.The circuit is based on Field Programmable Gate Array (FPGA) devices, then a set of specific purpose hardwaredesigns can be loaded on RAM cells in FPGA, so it's fast as hardware and versatile as software, and different approaches ofgradient can be implemented from host computer to programmable logic in FPGA.Keywords: moving images, time gradient, Fiel Programmable Gate Array, video rate, 3D mask, visual motion, imagesequences, hardware for image processing.