On minimum spanning tree streaming for hierarchical segmentation

Leonardo Gigli (1) , Santiago Velasco-Forero (1) , Beatriz Marcotegui (1)

1 CMM - Centre de Morphologie Mathématique

The minimum spanning tree (MST) is one the most popular data structure used to extract hierarchical information from images. This work addresses MST construction in streaming for images. First, we focus on the problem of computing a MST of the union of two graphs with a non-empty intersection. Then we show how our solution can be applied to streaming images. The proposed solution relies on the decomposition of the data in two parts. One stable that does not change in the future. This can be stocked or used for further treatments. The other unstable needs further information before becoming stable. The correctness of proposed algorithm has been proven and confirmed in the case of morphological segmentation of remote sensing images.

Domaines: Traitement du signal et de l'image [eess.SP]; Intelligence artificielle [cs.AI]; Mathématique discrète [cs.DM]; Traitement des images [eess.IV]; Machine Learning [stat.ML]; Applications [stat.AP]

Leonardo Gigli, Santiago Velasco-Forero, Beatriz Marcotegui. On minimum spanning tree streaming for hierarchical segmentation. Pattern Recognition Letters, 2020, 138, pp.155-162. ⟨10.1016/j.patrec.2020.07.006⟩. ⟨hal-02909738⟩