Most original work on feature extraction has its root in classical 2D image processing (Sec.1) and mainly focuses on edge detection and the localization of interest points and regions. In practice, extracting these features corresponds to segment the image and to analyze its content. The ad- vances in modeling techniques and the availability of effective 3D acquisition devices, like theodo- lites and LIDAR fly, led to a dramatic increase in the amount of 3D data available. Since images are not suited to represent all possible 3D data types, feature extraction algorithms have been specifically designed for understanding, filtering and organizing 3D data (Sec. 3.1 and 3.2). In this chapter, we discuss the main 3D approaches to the analysis of data in GIS, such as the anal- ysis of the evolution of level sets on terrain models and the identification of its ridges and ravines. Applications of the techniques discussed in Sec.3.2 to storm tracking and change detection will be presented in Sec.4.3