DeepTopoClustering: Unsupervised Derivation of Surface Process Taxonomy from 4D Point Clouds for Topographic Monitoring
DeepTopoClustering derives an unsupervised taxonomy of surface processes from 4D laser-scanning point clouds.
Permanent laser scanning yields high-frequency 4D point clouds, but existing methods poorly group detected surface activities into process types. DeepTopoClustering converts 4D objects-by-change into GeoMorphograms and learns embeddings with a convolutional autoencoder and hierarchical deep clustering. On two sandy-beach datasets and their combination, it reaches F1 0.78 and match accuracy 0.92 for eight major process types, outperforming dimensionality reduction and flat clustering.
- Unsupervised hierarchy learned from 4D objects-by-change on beach sites.
- GeoMorphograms encode the temporal evolution of bounded surface change.
- Eight process types reach F1 0.78 and match accuracy 0.92.
- Taxonomy separates erosion- and deposition-dominated activities and finer subtypes.
From the original article
1 min read at arxiv.org4D point clouds acquired by permanent laser scanning (PLS) enable accurate high-frequency monitoring of surface change in dynamic topographic environments. However, existing methods remain limited in organizing detected surface activities into meaningful process types.
By Jiapan Wang · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.