3. Producción
Browse
2 results
Search Results
- Some of the metrics are blocked by yourconsent settings
Item type:Publication, Dimensionality reduction via an orthogonal autoencoder approach for hyperspectral image classification(International Society for Photogrammetry and Remote Sensing, 2020-08-06)Abstract. Nowadays, the increasing amount of information provided by hyperspectral sensors requires optimal solutions to ease the subsequent analysis of the produced data. A common issue in this matter relates to the hyperspectral data representation for classification tasks. Existing approaches address the data representation problem by performing a dimensionality reduction over the original data. However, mining complementary features that reduce the redundancy from the multiple levels of hyperspectral images remains challenging. Thus, exploiting the representation power of neural networks based techniques becomes an attractive alternative in this matter. In this work, we propose a novel dimensionality reduction implementation for hyperspectral imaging based on autoencoders, ensuring the orthogonality among features to reduce the redundancy in hyperspectral data. The experiments conducted on the Pavia University, the Kennedy Space Center, and Botswana hyperspectral datasets evidence such representation power of our approach, leading to better classification performances compared to traditional hyperspectral dimensionality reduction algorithms. - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Development and validation of a novel metric for describing the three-dimensionality of bed forms(Elsevier, 2021-10-01)Description and characterization of bed forms in different environmental settings requires an assessment of three-dimensionality. This is related to the irregularity of the bed form shape and the degree of dissimilarity from strictly 2D bed forms, those bed forms that are straight, continuous, lack of defects and are arranged in groups with an aligned pattern. Three-dimensionality assessments are commonly vague and subjective, and even when some numeric relations for three-dimensionality quantification exist, they are based mainly on the characteristics of the bed form crest and cannot be easily implemented or automated to be applied to bed form fields. In this work an objective and general method is presented for the assessment of bed form three-dimensionality. The method introduces a three-dimensionality index Tb, which considers the correlation between bed profiles orthogonal to a target direction and a reference mean longitudinal bed profile. It is shown that the suggested method can serve as an objective characterization tool for ranking the three-dimensionality of groups of bed forms at different scales, with a continuous scale between 0 and 1. For archetypical bed forms, which were synthetically generated from sinusoidal functions, Tb successfully reproduced the intuitive hierarchy of the degree of three-dimensionality. Moreover, when applied to real subaqueous dunes in one-directional decelerating flows, Tb was able to reveal a gradual decrease of the bed configuration three-dimensionality with flow strength increase. The method also proved to be useful to automatically infer the orientation angle of bed forms by finding the angle of minimum three-dimensionality. This is exemplified with superimposed ripples on a Martian barchan-dune, using the grey scale of a high-resolution image instead of bed elevations. The new approach may be also useful to validate morphodynamic models, and to develop three-dimensionality maps for identifying heterogeneous regions or zones with a bad quality of survey data.