Decision Fusion With Multiple Spatial Supports by Conditional Random Fields

Devis Tuia*, Michele Volpi, Gabriele Moser

*Corresponding author for this work

Research output: Contribution to journalArticleAcademicpeer-review

11 Citations (Scopus)


Classification of remotely sensed images into land cover or land use is highly dependent on geographical information at least at two levels. First, land cover classes are observed in a spatially smooth domain separated by sharp region boundaries. Second, land classes and observation scale are also tightly intertwined: they tend to be consistent within areas of homogeneous appearance, or regions, in the sense that all pixels within a roof should be classified as roof, independently on the spatial support used for the classification. In this paper, we follow these two observations and encode them as priors in an energy minimization framework based on conditional random fields (CRFs), where classification results obtained at pixel and region levels are probabilistically fused. The aim is to enforce the final maps to be consistent not only in their own spatial supports (pixel and region) but also across supports, i.e., by getting the predictions on the pixel lattice and on the set of regions to agree. To this end, we define an energy function with three terms: 1) a data term for the individual elements in each support (support-specific nodes); 2) spatial regularization terms in a neighborhood for each of the supports (support-specific edges); and 3) a regularization term between individual pixels and the region containing each of them (intersupports edges). We utilize these priors in a unified energy minimization problem that can be optimized by standard solvers. The proposed 2L$łightning$CRF model consists of a CRF defined over a bipartite graph, i.e., two interconnected layers within a single graph accounting for interlattice connections. 2L$łightning$CRF is tested on two very high-resolution data sets involving submetric satellite and subdecimeter aerial data. In all cases, 2L$łightning$CRF improves the result obtained by the independent base model (either random forests or convolutional neural networks) and by standard CRF models enforcing smoothness in the spatial domain.
Original languageEnglish
Pages (from-to)3277-3289
JournalIEEE Transactions on Geoscience and Remote Sensing
Issue number6
Early online date17 Mar 2018
Publication statusPublished - 2018


  • Classification
  • conditional random fields (CRF)
  • convolutional neural networks (CNNs)
  • hierarchical models
  • region-based analysis
  • semantic labeling.

Fingerprint Dive into the research topics of 'Decision Fusion With Multiple Spatial Supports by Conditional Random Fields'. Together they form a unique fingerprint.

Cite this