A case study for a multitemporal segmentation approach in optical remote sensing images.

Continuous observations from remote sensors provide high temporal and spatial resolution imagery, and better remote sensing image segmentation techniques are mandatory for efficient analysis. Among them, one of the most applied segmentation techniques is the region growing algorithm. Within this context, this paper describes a study case for a multitemporal segmentation that adapts the traditional region growing technique. Our method aims to detect homogeneous regions in space and time observing a sequence of optical remote sensing images. Tests were conducted by considering the Dynamic Time Warping distance as the homogeneity criterion to grow regions. A case study on high temporal resolution for sequences of Landsat-8 vegetation indices products provided satisfactory outputs.GEOProcessing 2018