Homework 1
CS 591Q/791V - Pattern Recognition
Instructor: Dr. Arun Ross
Due Date: Feb 7, 2008
Total Points: 40
Note: You are permitted to discuss the following questions with others in the class.
However, you must write up your own solutions to these questions. Any indication to the
contrary will be considered an act of academic dishonesty.
1. [20 points] The paper Veggie Vision by Bolle et al. discusses a pattern recognition system
that automatically classifies produce items at the checkout counter of a grocery store. Briefly
describe this system based on the pattern recognition terminology developed in class: (i)
sensors deployed; (ii) segmentation algorithm; (iii) features used; (iv) classification scheme;
and (v) post-processing methodology (see Figure 1.7 in DHS). How is classifier training
accomplished by Veggie Vision?
2. [20 points] Chapter 1 of the book introduces an example of a pattern classification system
that distinguishes sea-bass from salmon. Present a different example of a pattern classifi-
cation system which may be of some commercial use. Choose a classification problem for
which you can actually collect a few training samples (∼5 samples per class) and measure
features on them. Briefly describe the (i) pattern classes, (ii) transducer to be used, (iii)
preprocessing and segmentation routines, and (iv) features that may separate these two
classes. Why would you be interested in automating this classification problem.
Select two important features for this problem and show the 1-dimensional histograms (class-
conditional densities) for these two features (see Figures 1.2 and 1.3 of the book). Also show
the 2-dimensional scatter plot when both features are used simultaneously. Are the pattern
classes reasonably well-separated in the 2D feature space that you have chosen?