Scalable Bag of Subpaths Kernel for Learning on Hierarchical Image Representations and Multi-Source Remote Sensing Data Classification

Abstract

Abstract:The geographic object-based image analysis (GEOBIA) framework has gained increasing interest for the last decade. One of its key advantages is the hierarchical representation of an image, where object topological features can be extracted and modeled in the form of structured data. We thus propose to use a structured kernel relying on the concept of bag of subpaths to directly cope with such features. The kernel can be approximated using random Fourier features, allowing it to be applied on a large structure size (the number of objects in the structured data) and large volumes of data (the number of pixels or regions for training). With the so-called scalable bag of subpaths kernel (SBoSK), we also introduce a novel multi-source classification approach performing machine learning directly on a hierarchical image representation built from two images at different resolutions under the GEOBIA framework. Experiments run on an urban classification task show that the proposed approach run on a single image improves the classification overall accuracy in comparison with conventional approaches from 2% to 5% depending on the training set size and that fusing two images allows a supplementary 4% accuracy gain. Additional evaluations on public available large-scale datasets illustrate further the potential of SBoSK, with overall accuracy rates improvement ranging from 1% to 11% depending on the considered setup.

Deep Learning for Classification of Hyperspectral Data: A Comparative Review

“…In particular, [33] introduced the possibility to design spatial-spectral kernels for SVMs able to handle hyperspectral data. This technique will then be largely adopted in later works [34], [35], [36]. With a similar objective, [37] proposes a methood to choose automatically the filters which lead to the most efficient features for hyperspectral data classification from a random-filter bank.…”

Adversarial Learning Based Discriminative Domain Adaptation for Geospatial Image Analysis

“…All of these methods for hyperspectral image analysis focus on spectral signatures only and do not take the spatial features into account. Spatial features are useful in understanding the semantics of the particular scene which can improve model performance [39], [40], [41] for hyperspectral image analysis.…”

Watershed-Based Attribute Profiles With Semantic Prior Knowledge for Remote Sensing Image Analysis

“…For the building extraction task, all classes except buildings are merged into a single background class. Following previous works [35], [34], training pixels were extracted from the first fifteen images of this dataset, and the remaining five images (zh16, zh17, zh18, zh19 and zh20) were used for evaluation. As for the Vaihingen dataset, the background class of the Zurich dataset is not considered for land-cover classification.…”