Implementation of image processing applications with evolutionary fault recovery scheme

This paper presents an evolutionary fault recovery scheme using Evolvable Hardware (EHW) to have reliable computations. The scheme is based on Genetic Algorithm as Evolutionary Algorithm used in EHW and uses Look Up Tables (LUTs) as basic reconfigurable elements. The scheme can recover a component with more than one fault and uses limited spare cells. The recovered hardware shows reasonable functionality when used in image processing applications. A 4-bit comparator, which is injected with up to 12 faults, is tested for an image median filtering. After recovery, there is only l.51dB loss in PNSR of the considered median filter.