Fundamentals of Image Processing

Abstract

In this chapter, the fundamentals that are necessary for the following chapters are introduced. A comprehensive introduction to the basics of image processing is given in [Azad et al., 2008].

The utilized camera model is introduced in Section 4.1. Common segmentation techniques for robotic vision applications are briefly summarized in Section 4.2. In Section 4.3, correlation techniques are introduced, which are used by the proposed object recognition systems. Homographies and affine transformations are briefly introduced in Section 4.4, including least-square methods for their computation on the basis of 2D-2D point correspondences. A brief introduction to principal component analysis is given in Section 4.5, which is applied for compression of the views in the proposed shape-based object recognition and pose estimation approach. In Section 4.6, the concept of particle filtering is explained, which forms the statistical framework for the proposed human motion capture system. The RANSAC method, which is used in the presented systems for filtering data sets before application of least squares methods, is explained in Section 4.7.