Classification non supervis{é}es d'acquisitions hyperspectrales cod{é}es : quelles v{é}rit{é}s terrain ?

2025-08-04

AI Summary

This work addresses inherent limitations of ground truth (GT) in unsupervised hyperspectral image classification—namely, ambiguous class definitions, substantial intra-class spectral variability, and significant labeling errors. Leveraging DD-CASSI–based compressive hyperspectral data acquisition, we propose a streamlined evaluation framework centered on modeling intra-class spectral variability. This framework exposes the unreliability of conventional GT for unsupervised assessment and advocates a paradigm shift toward spectral-consistency–driven evaluation. Our method integrates DD-CASSI sensing, unsupervised clustering, and spectral consistency analysis. Evaluated on the Pavia University dataset, it demonstrates robust identification of spectrally coherent land-cover regions even at a 10× compression ratio. Consequently, it substantially enhances the physical interpretability of classification results and improves the fidelity of authenticity assessment—without reliance on imperfect GT annotations.

Technology Category

• Computer Vision: Remote Sensing / Geospatial AI
• Machine Learning: Calibration & Uncertainty Quantification
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Web Mining and Content Analysis: Normalization, clustering, classification, and summarization of Web text

Abstract

We propose an unsupervised classification method using a limited number of coded acquisitions from a DD-CASSI hyperspectral imager. Based on a simple model of intra-class spectral variability, this approach allow to identify classes and estimate reference spectra, despite data compression by a factor of ten. Here, we highlight the limitations of the ground truths commonly used to evaluate this type of method: lack of a clear definition of the notion of class, high intra-class variability, and even classification errors. Using the Pavia University scene, we show that with simple assumptions, it is possible to detect regions that are spectrally more coherent, highlighting the need to rethink the evaluation of classification methods, particularly in unsupervised scenarios.

Problem

• Unsupervised classification of compressed hyperspectral data
• Addressing limitations in ground truth evaluation methods
• Improving spectral coherence in unsupervised classification

Innovation

• Unsupervised classification with DD-CASSI imager
• Estimates reference spectra despite compression
• Detects spectrally coherent regions simply