Surface Displacement Measurement from Remote Sensing Images

2Image Matching and Optical Sensors

Marc PIERROT DESEILLIGNY and Ewelina RUPNIK

LaSTIG, ENSG-IGN, Gustave Eiffel University, Saint-Mande, France

2.1. Introduction, definition and applications

Image matching is a well-known problem in the Earth sciences and features in two main applications: 1) the computation of 3D geometry from images acquired from different viewpoints and 2) the computation of surface deformations from images acquired from similar points of view. Many scientific communities (photogrammetry, robotics, computer vision, Earth science, etc.) have investigated image matching problems, which has led to many different naming conventions. In this chapter, we will interchangeably use the terms image matching and image correlation to describe the same concept, which is introduced in the following section. The image matching concepts and methods that are described in this chapter are presented in the context of optical data. Most of them can also be applied to SAR amplitude images. However, since the radiometry and geometry of SAR sensors are different from optical sensors, methods specific to SAR imagery will be presented in Chapter 3.

2.1.1. Problem definition

Given a pair of images, let us assume that for some physical reasons, one image is a deformation of the other. The goal of image matching is to recover this deformation, given the images. In mathematical terms, let I₁ and I₂ be the images we want to match, and let I₂ be the deformation of I₁ by a certain geometric ...