A FPGA implementation of low-complexity noise removal

Impulse noise removal is a very important preprocessing operation in many computer vision applications. This paper presents a low-complexity noise removal based on a conditional technique and outlines its FPGA implementation for window sizes of (3×3) and (5×5). As the experiments show, the proposed technique performs significantly better than standard median filter and achieves superior image quality. The FPGA implementations are very compact, fast and consume low-power.