Analysis and Calibration of a Mirror Setup producing Mirror-reflected, Multi-view Videos

5 Chapter 2 Literature Review In this chapter the literature review is reported. We start by talking about the basic principles related to image processing, then we will talk about the most important 3D imaging systems, and some of the automatic evaluation methods that exist for the assessment of progress of the patient about the facial paralysis. The comparison between the existing 3D imaging systems will take us to outline the advantages and disadvantages of these systems, and then to motivate the use of SMILE system proposed by Frey et al., 1999. The problems about the used SMILE system will take us to introduce a new cylindrical calibration device. In this regard, in the later section, we will talk about the most important calibration algorithm, checkerboard pattern and corner detector, camera model and how this is used to map more 2D points to a single 3D point. 2.1 Image Processing Computer vision is the discipline that tries to emulate human vision not only as the acquisition of a two-dimensional image but also to enable machines to understand and interpret the visual information obtained from 2D still images or video sequences. Computer Vision is a rapidly growing field thanks to the availability of more accurate and less expensive cameras as well as the high computational power of modern computers. Computer vision has a very wide scope, the camera is, in fact, a non- invasive tool that can operate at high distances if equipped with appropriate optics. The versatility of computer vision is due to the great amount of information carried by an image, and the ability to extract only that information that is necessary for the purpose of the application, while the accuracy of the results can be strongly influenced by factors ranging from camera resolution to algorithms configuration. An artificial vision system is usually made of one or more cameras connected to a computer, this is used to automatically interpret the images of a real scene, and so to get useful information for the application. What has been said up to now can be summarized in the following work flow. FIGURE 2.1: Image processing workflow There are three types of operations that can be performed on an image.