Change Detection and Image Time-Series Analysis 1

3An Overview of Covariance-based Change Detection Methodologies in Multivariate SAR Image Time Series

Ammar MIAN^{1, 2}, Guillaume GINOLHAC², Jean-Philippe OVARLEZ³, Arnaud BRELOY⁴ and Frédéric PASCAL¹

¹CentraleSupélec, Paris-Saclay University, Gif-sur-Yvette, France

²University Savoie Mont Blanc, Annecy, France

³ONERA, Palaiseau, France

⁴Paris Nanterre University, France

3.1. Introduction

Change detection (CD) for remotely sensed images of the Earth has been a popular subject of study in the past decades. It has indeed attracted a plethora of scholars due to the various applications, in both military (activity monitoring) and civil (geophysics, disaster assessment, etc.) contexts. With the increase in the number of spatial missions with embedded synthetic aperture radar (SAR) sensors, the amount of readily available observations has now reached the “big data” era. To efficiently process and analyze this data, automatic algorithms therefore have to be developed. Notably, CD algorithms have been thoroughly investigated: the literature on the subject is dense, and a variety of methodologies can be envisioned¹.

Figure 3.1. General procedure for a change detection methodology

Broadly speaking, a change detection algorithm can be synthesized as in Figure 3.1, and it relies on three main separate elements:

• – a pre-processing phase, in which the time series of images have to be co-registered, ...