Texture Analysis Filters
The Texture Analysis filters can be used to characterize the textures within an image.
| What it does | Settings | |
|---|---|---|
| Gabor |
Gabor filters, which are often used in texture analysis, edge detection, feature extraction, and other applications, are special classes of bandpass filters that allow a certain 'band' of frequencies and reject the others.
As shown below, a Gabor filter gives the highest response at edges and at points where texture changes. The following images show an image and its transformation after the filter is applied with varying values for Theta (0, 45, 90). Other settings selected (from left to right) were; Kernel size = 21, Sigma = 10.0, Wavelength = 10.0, Spatial aspect ratio = 0.05.
References |
Kernel size
Sigma Theta Wavelength Spatial aspect ratio |
| Image Moments |
As shown in the example below, this filter calculates all image moments up to a certain order.
|
Kernel size
Order |
| Local Binary Pattern | The local binary pattern filter provides a simple, yet effective method for representing the texture of an image by encoding the local spatial relationship between the pixels.
Each pixel in an image is compared to its surrounding pixels to determine if the surrounding points are greater than or less than the central point. The result of these comparisons is a binary pattern that encodes the local texture of the image, as shown below.
|
Angular count
Radius Method |
| Membrane Projections | As sown below, this filter enhances membrane-like structures of an image through directional filtering.
|
Kernel size
Projection type Number of angles |
