Work

Assessment of the Information Content of Vegetation Indices Set for Hyperspectral Data Preprocessing

Year: 2026

Type: conference-paper

Abstract: An approach to assessing the informativeness of vegetation indices and their combinations with principal components for hyperspectral data preprocessing is proposed. It combines separability metrics, area under the curve, and Jeffries–Matusita distance with a Spearman's correlation threshold to generate compact, selective feature sets. Using the proposed algorithm with a 2D convolutional neural network yielded classification accuracy comparable to or exceeding principal component analysis, while preserving input-feature interpretability.

DOI: 10.1109/et72079.2026.11768360

Authors: Danila A. Titov, Victoria A. Ryzhova, Sergey N. Yarishev, Todor S. Djamiykov, Marin B. Marinov

Institutions: ITMO University, Technical University of Sofia

Language: English

Topic: Remote-Sensing Image Classification

Subfield: Computer Vision and Pattern Recognition

Field: Computer Science

Domain: Physical Sciences

Open Access status: gold

OpenAlex ID: W7225699286

FWCI:

Cites: 7

Cited by:

Related to: 10