Bookmark and Share

Computer Vision Metrics: Chapter Four (Part F)

Register or sign in to access the Embedded Vision Academy's free technical training content.

The training materials provided by the Embedded Vision Academy are offered free of charge to everyone. All we ask in return is that you register, and tell us a little about yourself so that we can understand a bit about our audience. As detailed in our Privacy Policy, we will not share your registration information, nor contact you, except with your consent.

Registration is free and takes less than one minute. Click here to register, and get full access to the Embedded Vision Academy's unique technical training content.

If you've already registered, click here to sign in.

See a sample of this page's content below:

For Part E of Chapter Four, please click here.

Bibliography references are set off with brackets, i.e. "[XXX]". For the corresponding bibliography entries, please click here.

Computer Vision, Models, Organization

This section contains a high- level overview of selected examples to illustrate how feature metrics are used within computer vision systems. Here, we explore how features are selected, learned, associated together to describe real objects, classified for efficient searching and matching, and used in computer vision pipelines. This section introduces machine learning, but only at a high level using selected examples. A good reference on machine learning is found in [546] by Prince. A good reference for computer vision models, organization, applications, and algorithms is found in Szelinski [324].

Several terms are chosen and defined in this section for the discussion of computer vision models, namely feature space, object models, and constraints. The main topics for this section include:

  • Feature spaces and selection of optimal features
  • Object recognition via object models containing features and constraints
  • Classification and clustering methods to optimize pattern matching
  • Training and learning

Note: Many of the methods discussed in computer vision journals and courses are borrowed from other tangent fields and applied, for example, machine learning and statistical analysis. in some cases computer vision is driving the research in such tangent fields. since these fields are well established and considered beyond the scope of this work, we provide only a brief topical introduction here, with references for completeness [546,324].

Feature Space

The collection and organization of all features, attributes, and other information necessary to describe objects may be called the feature space. Features are typically organized and classified into a feature space during a training or learning phase using ground truth data as a training set. The selected features are organized and structured in a database or a set of data structures, such as trees and lists, to allow for rapid search and feature matching at run time.

The feature space may contain one or more types of descriptors using spectra such as histograms, binary pattern vectors, as multivariate composite descriptors. In addition, the feature space contains constraints used to associate sets of features together to identify objects and classes of objects. A feature space is unique to any given...