FPGA-based Pipeline Architecture toTransform Cartesian Imagesinto Foveal ImagesbyUsing a new Foveation Approach

InVision systems theimage processing represents abottle neckbecause thebigamountofinformation thatshouldbeanalyzed. Working withvariant spacesoverthevisual field hasbeenwidely proposed asa waytoreduce suchinformation. Foveal vision isoneofthese proposals byproviding a waytotransform thevisual field obtained with conventional camerasintoa sampling withhigh resolution atthecenter anddecreasing overthe periphery suchasinmammalvision systems. Inthis paper, anFPGA basedarchitecture totransform conventional images into foveal images ispresented. Thehardware algorithm hasbeentaken fromnew proposal tofoveate images. Strategies asparallelism andpipeline areexploited toobtain a high performance andthus, withbothofthem, reduction inthevisual field andthetransformation inreal timeofthedigital imagesintofoveal images, a vision system canaccelerate itsperformance and reaches real timerestrictions.