Digital Architecture for a Median Filter of Image Based on Sorting Network

摘要

In many image processing applications, one of the most frequent problems is the existence of impulsive noise. Median filtering has been proved to be a very powerful method to overcome that problem. However, it results to be computationally intensive because of its requirement of sorting data. In this paper, a digital architecture for median filtering of grayscale images, which is based on a nine-data sorting network, is presented. The functional behavior of the architecture is verified by Matlab programming and the hardware implementation tested on a Spartan-3E FPGA device.