COMP7702 Pattern Recognition

Selected modern pattern recognition techniques will be covered in depth, including segmentation techniques such as threshold-based methods or graph-based methods, feature description and extraction (intensity features, shape features, principle component analysis), feature selection techniques, classification methods and evaluation methodology such as ROC curve, sensitivity and specificity, Dice and Jaccard index.

This topic aims to provide:

• An understanding of pattern recognition problems in digital images
• An introduction to computer methods for pattern recognition problems
• Experience in scientific computing
• Experience in applying different computer methods in solving a range of real-life examples of pattern recognition problems
• Skills necessary to analyse, solve and critically evaluate, using appropriate evidence, feasibility of various approaches to a wide range of pattern recognition problems

1. Understand the foundations of selected pattern recognition techniques
2. Understand the concept and theory behind a range of common techniques in solving a broad range of pattern recognition problems
3. Analyse and explain issues arising in real-life pattern recognition applications
4. Use software to solve new pattern recognition problems
5. Critically evaluate applicability of various techniques to new problems
6. Use appropriately sourced evidence to devise feasible solutions for complex image analysis tasks