Invariant texture segmentation via circular Gabor filters

Abstract

In this paper, we focus on invariant texture segmentation, and propose a new method using circular Gabor filters (CGF)

E2CAV, Pavement layer thickness estimation system based on image texture operators

“…The second method available in the software tool for texture segmentation is Gabor filters. These filters are special cases of band pass filters, so they are commonly used in edge detection, stereo images disparity estimation and feature extraction (Zhang, Tan, & Ma, 2002). In this work, the Gabor filter used is defined in equation (6).…”

“…Where x and y stand for the pixel coordinates, g denotes the spatial aspect ratio, s denotes the standard deviation of the Gaussian function, l stands for the wavelength of the sinusoidal factor and f denotes the phase offset. Gabor filters are parameterized using 4 variables namely g, s, l and f. According with (Salman et al, 2013) and (Zhang et al, 2002) the phase offset f is divided in six different angles between 0 and p as depicted in equation (7).…”

“…The wavelength of the sinusoidal factor l depends on the image number of columns (Zhang et al, 2002) since it has a direct relationship with the frequency domain. Then, different cut-off frequencies are selected according to equation (8).…”

“…. In (Sandler & Lindenbaum, 2006) and (Zhang et al, 2002), the authors propose a relationship between s and l which is described in equation (10).…”

Efficient design of adjustable circular IIR 2D filters

“…Using appropriate filter banks, the frequency spectrum of the texture image is decomposed into a number of sub-bands. An important application is texture segmentation and classification [11]- [13], in particular using rotation invariant filters. A class of efficient Gaussian-shaped circular filters implemented in cellular neural networks was approached in [16].…”