We present an approach for segmentation of 3D voxel data based on algebraic topology. The framework leads to a single pass algorithm for constructing symbolic descriptions of volumertic data. Unlike previous algorithms segmentation is not limited to sxtracting isosurfaces, i.e. 2D manifolds, enabling the processing of image sequences. Due to the anchoring on algebraic topology, the correctioness of the algorithm can be proven.
This paper presents the change detection analysis of two multispectral datasets for the Bostanlik District of Tashkent, Uzbekistan, using Landsat-5 TM data for 1989 and Landsat-8 OLI for 2017. Both supervised classification and maximum likelihood algorithms were utilized for the change detection analysis. Six land use classes were identified: snow cover, bare soil/rock, forest, waterbody, built-up areas and agriculture. The change detection technique showed that within 28 years, significant changes occurred in the classes of the forest, built-up areas, bare soil and snow cover. The presented results might be valuable for the government authorities and stakeholders for future land use planning activities.