ABSTRACT
To fuse two sensor readings, they must be in a common coordinate system. The assumption that the mapping is known a priori is unwarranted in many dynamic systems. Finding the correct mapping of one image onto another is known as registration. An image can be thought of as a two-dimensional (2D) sensor reading. In this chapter we will provide examples using 2- and 2½D data. The same approaches can be trivially applied to 1D readings. Their application to data of higher dimensions is limited by the problem of occlusion, where data in the environment is obscured by the relative position of objects in the environment. The first step in fusing multiple sensor readings is registering the images to find the correspondence between them [Gonzalez 92].
