Classification, Parameter Estimation and State Estimation, 2nd Edition

9Worked Out Examples

In this chapter, three simple examples on real image data will be presented. It is the intention that readers who are new to PRTools just get some flavour about building and applying a recognition system. Then three worked out examples of the topics discussed in this book will be given: classification, parameter estimation and state estimation. They will take the form of a step-by-step analysis of datasets obtained from real-world applications. The examples demonstrate the techniques treated in the previous chapters. Furthermore, they are meant to illustrate the standard approach to solve these types of problems. Obviously, the Matlab^® and PRTools algorithms presented in the previous chapters will be used. The datasets used here are available through the website accompanying this book and www.37steps.com.

9.1 Example on Image Classification with PRTools

9.1.1 Example on Image Classification

In this section an example on real-world colour images will be presented. We will go through the program step by step. If possible, copy the lines in Matlab^® and look at the response in the command window or in a graphical window elsewhere on the screen. This example deals with a set of 256 colour images, the Delft Image Database (delft_idb) stored in a datafile (see Figure 9.1). They are organized in nine classes named books, chess, coffee, computers, mug, phone, plant, rubbish and sugar. First it should be checked to see whether the data are available:

prdatafiles ...