A Low-Complexity and Low Power Design of 2D-Median Filter

Main Article Content

Takeaki Matsubara
Vasily G. Moshnyaga
Koji Hashimoto

Abstract

Impulse noise removal is a very important preprocessing operation in many computer vision applications. Usually it is accomplished by median filter with excessive sorting and therefore large power. This paper presents a new design of 2D median filter that utilizes a simple conditional filtering technique, executes fewer computations than related designs while achieving superior image quality. Experimental FPGA implementation of the proposed filtering scheme is compact, fast and low-power consuming.

Article Details

How to Cite
[1]
T. Matsubara, V. G. Moshnyaga, and K. Hashimoto, “A Low-Complexity and Low Power Design of 2D-Median Filter”, ECTI-CIT Transactions, vol. 5, no. 2, pp. 89–97, Apr. 2016.
Section
Artificial Intelligence and Machine Learning (AI)