Lab 1. Introduction

Outline

1. Introduction (about 10 minutes)
2. STPR toolbox installation (15 minutes)
3. Refreshing your knowledge of Matlab (15 minutes)
4. Simple data task (up to 55 minutes)

STPR toolbox installation

Expected time duration: 15 minutes
Statistical Pattern Recognition Toolbox contains variety of algorithms which you will implement yourself during this course (you can use them as an inspiration) and contains also many support functions which we will use throughout the semester.

Download the latest version of the toolbox from its pages. Unpack the content of the .zip file into your working directory (you should be able to find it next week!). To be able to use the toolbox, you have to set the path to it:

1. Run Matlab
2. Change the path to the directory where you unpacked the toolbox (use cd command or the path combo box at right top of the Matlab window)
3. Running stprpath will set the path and do initialization of the toolbox

Now you can switch to any other directory and still use the functions from the STPR toolbox.

To verify the success of previous steps try to run demo_ocr.

Refreshing you knowledge of Matlab

Expected time duration: 15 minutes
We will be using Matlab programming language during whole semester. It is necessary for you to be as familiar with it as possible, since the labs are time limited and the tasks are often challenging. It may happen that you spend more time looking for Matlab syntax of some function instead of solving the problem itself.

For the case you are not too sure about your Matlab skills, here are few useful links:

• Very short revision of basic operations.

• Getting started with Matlab. Nicely structured tutorial from Indiana University. PDF and HTML version.

• MIT tutorial. Compact, concise tutorial. Describes work with Matlab under Unix but most of it is applicable to any operating system (just skip the section "Starting a Session")

• Programming in MATLAB. PDF document suitable for printing. Part of tutorials page at Southern Illinois University in Carbondale.

As part of today’s assignment, do following (avoiding the use of any loops in your program):

1. Create an arbitrary 4x4 matrix A and find its transpose,
2. Print out third column of the matrix A, print out last two rows from second to the last column of the matrix,
3. Find all positions in A greater then 3 and increment them by 1,
4. Add a column of ones to the matrix A.

Simple data task

Expected time duration: up to 55 minutes

In this part of the exercise, we will work with a simple input data and we will split it into training and test part. The data contains images of letters. Do the following:

1. Download the data file with the letter images and save it to your working directory,
2. Load the file to Matlab (function load). You will get variables images (3D array of images), Alphabet (letters in the images) and labels (indexes of the images into Alphabet array),
3. Display one or more images for each letter in the Alphabet list (see doc on functions imshow, imagesc, subplot, montage),

Try this code:

ims = reshape(images,size(images,1),size(images,2),1,size(images,3));
montage(ims); colormap gray;

4. Compute the mean images of you name initials and display them (You will probably encounter a problem with non-decimal numbers in the image matrix. You can avoid them by using floor or uint8 function),
5. For all letters plot the histograms of dark pixels (value below 128) in the image. Plot all the histograms into one graph to compare them.

---

Created by Jan Šochman, 4.10.2006, last update 22.2.2008