Egomotion and Related Tasks
Humans employ 3-D vision to find their way around in the process called egomotion. The necessary visual processes are carried out with ease and very much in real time. What shortcuts are being taken, and what cues does the brain use to help make the necessary rapid analysis? This chapter considers these issues and finds some interesting methodologies.
• how vanishing points can be found and used to help provide a basic understanding of the scene.
• how a plan view of the ground plane can be obtained and used to help with robot navigation.
• how known measurements of flagstones or other objects can be used to help identify the ground plane and locate the vanishing points.
• how perspective distortions of object center locations may be eliminated.
• how vehicles can be guided using vision to compensate for roll, pitch, and yaw.
This chapter ignores camera calibration and the ultimate need to cope with general 3-D transformations using homogeneous coordinates, both of which are studied in detail in Chapter 21. They have been left until last in order not to demotivate readers who are more orientated toward applications.