Image Processing Reference
In-Depth Information
8
Introduction to texture
description,
segmentation and
classification
8.1
Overview
This chapter is concerned with how we can use many of the feature extraction and description
techniques presented earlier to characterise regions in an image. The aim here is to describe
how we can collect measurements for purposes of recognition, using texture.
We shall first look at what is meant by texture and then how we can use Fourier
transform techniques, statistics and region measures to describe it. We shall then look at
how the measurements provided by these techniques, the description of the texture, can be
collected together to recognise it. Finally, we shall label an image according to the texture
found within it, to give a segmentation into classes known to exist within the image. Since
we could be recognising shapes described by Fourier descriptors, or by region measures,
the material is actually general and could be applied for purposes of recognition to measures
other than texture.
Table 8.1
Overview of Chapter 8
Fourier Transform
Energy
Texture description
Co-occurrence
Entropy
Regions
Inertia
Texture classification
k nearest neighbour rule
Convolution
Texture segmentation
Tiling
Thresholding
 
 
Search WWH ::




Custom Search