Feature Profiles from Attribute Filtering for Classification of Remote Sensing Images

Abstract

International audienc

Local Feature-Based Attribute Profiles for Optical Remote Sensing Image Classification

“…For attribute filtering, we exploited two attributes including the area and the moment of inertia. Ten area thresholds were adopted for the Reykjavik data as proposed by several papers [20], [39], [40]. For the Pavia University data, fourteen thresholds were automatically computed according to [24].…”

Classification of Remote Sensing Images Using Attribute Profiles and Feature Profiles from Different Trees: A Comparative Study

“…We have revisited the principles of APs [1] and FPs [3] with the aim to conduct a comparative study of their performance on remote sensing image classification. Our experiments have taken into account various tree structures for their generation including the component, in- clusion and partition trees.…”

“…In order to better characterize the region/object enclosed by the CC (i.e. which corresponds to a filtered tree's node), node features are extracted instead of the node's gray level in the recently proposed FPs [3] . Figure 1 provides an overview of how the generation of FPs differs from the standard AP technique.…”

“…Here, standard deviation and area were selected thanks to their stable performance from lots of our experiments. The attribute threshold values were set as in [3,6,7]. For the hyperspectral Pavia image, the first four PCA components were exploited as in most of related work.…”

Attribute Profiles For Satellite Image Time Series

“…A major drawback of such a strategy is the difficulty to reconstruct the filtered image (or SITS here) since there could be several different vectors (or time series) being considered as equivalent and thus it is not possible to choose among these ties if they correspond to the retained maxima or minima. Nevertheless, we can avoid the reconstruction phase, similarly to the concept of feature profiles (FP) [10].…”