Visual pattern recognition
What Will We Learn?
- What is visual pattern recognition and how does it relate to general pattern recognition?
- What are patterns and pattern classes?
- What is a pattern classifier?
- Which steps are normally needed to design, build, and test a visual pattern classifier?
- How can the performance of visual pattern classifiers be evaluated?
This chapter presents the basic concepts of pattern recognition (also known as pattern classification) and introduces a few representative techniques used in computer vision. These techniques assume that an image has been acquired and processed and its contents have been represented using one or more of the techniques described in Chapter 18. The goal of pattern classification techniques is to assign a class to each image (or object within an image) based on a numerical representation of the image's (or object's) properties that is most suitable for the problem at hand.
Pattern classification techniques are usually classified into two main groups: statistical and structural (or syntactic). In this chapter, we exclusively focus on statistical pattern recognition techniques, which assume that each object or class can be represented as a feature vector and make decisions on which class to assign to a certain pattern based on distance calculations or probabilistic models. Since the techniques presented in this chapter work with numerical feature vectors, regardless of the meaning of their contents ...