This chapter discusses the mapping of real world objects to the image plane, including the geometric and the radiometric parts of image formation. The image data and prior knowledge on the considered application are the major sources of information for various vision issues. Common examples are image restoration, filtering, segmentation, reconstruction, modeling, detection, and recognition or pose estimation algorithms. The hardest problems in computer vision are related to object recognition. There has been no general algorithm that allows the automatic learning of arbitrary 3-D objects and their recognition and localization in complex scenes. State-of-the-art approaches dealing with high-level vision tasks are essentially dominated by model-based object recognition methods. Most techniques apply intuitive ideas specified for the given application and neglect the exploration of precisely defined mathematical models. There exists no unified theoretical formalization that provides the framework for the analytical analysis of designed complete systems.
