Segmentation of forest stands in very high resolution stereo data

Classification of forest stands is an important task in forest inventory and monitoring. The multi view angle data of the new digital stereo line scanner HRSC-A was used in this work to investigate the applicability of semivariogram textural parameters for a stand segmentation of variable aged spruce stands based on viewing angle related textural features. Results indicate a potential of directional semivariogram parameter in stand structure differentiation, especially when homogeneous stands are examined. Sensor viewing angles are used and Sun illumination angle is taken into account to increase the separability of texture features.