Comparative Analysis of Filtering Techniques on Images of Building Structures with Cracks

Abstract

The presence of noise causes image distortion which can be
eliminated by image filtering. In recent years, it has been proven that
noise has a significant effect on the accuracy of performance
evaluation which has motivated many researchers in the field of
image processing and computer vision to use various filtering
techniques. This paper therefore looked into the three most used
filters (Averaging, Median and Wiener filters) by researchers in this
field so as to have a comprehensive comparative analysis of the
filters on images of building structures with cracks. These three
different filters were applied on 100 images of building structure with
cracks, after which thresholding operation was performed in order
to segment the crack from the background image. The Area Under
Curve (AUC) shows that the median filter outperformed the other
filters. The median filter value is 0.77778, Averaging filter value is
0.77641 and Wiener filter value is 0.77503