Pit Pattern Classification in Colonoscopy using Wavelets - WaveLab

4 Automated pit pattern
classification
4.1 Introduction
In order to succeed, your desire for success
should be greater than your fear of failure.
- Bill Cosby
Based on the preliminaries introduced in the last two chapters this chapter now presents
methods we developed for an automated classification of pit pattern images. We describe
a few techniques and algorithms to extract features from image data and how to perform a
classification based on these features. Additionally this chapter introduces two classification
schemes without any wavelet subband feature extraction involved.
4.2 Classification based on features
The methods presented in this section first extract feature vectors from the images used,
which are then used to train classifiers. The used classifiers and the process of classifier
training and classification is then further described in section 4.4.
4.2.1 Feature extraction
4.2.1.1 Best-basis method (BB)
In section 2.6.1 two different methods for basis selection have been presented. The classification
scheme presented in this section is based on the best-basis algorithm, which, as
already stated before, is primarily used for compression of image data. This approach however
uses the best basis algorithm to try to classify image data.
During the training process all images in I T are decomposed to obtain the wavelet packet
coefficients for each image. Additionally during the decomposition some cost information
is stored along with each node of the respective quadtree. This cost information is based
on the chosen cost function and is used to determine the importance of a subband for the
35