Published September 20, 2007 | Version 4284

A Technique for Improving the Performance of Median Smoothers at the Corners Characterized by Low Order Polynomials

Authors/Creators

Description

Median filters with larger windows offer greater smoothing and are more robust than the median filters of smaller windows. However, the larger median smoothers (the median filters with the larger windows) fail to track low order polynomial trends in the signals. Due to this, constant regions are produced at the signal corners, leading to the loss of fine details. In this paper, an algorithm, which combines the ability of the 3-point median smoother in preserving the low order polynomial trends and the superior noise filtering characteristics of the larger median smoother, is introduced. The proposed algorithm (called the combiner algorithm in this paper) is evaluated for its performance on a test image corrupted with different types of noise and the results obtained are included.

Files

4284.pdf

Files (588.5 kB)

Name Size
md5:23286a522c27510e924c4e6140a93f7d
588.5 kB Preview Download