[Spectral Uncertainty of Terrestrial Objects and the Applicability of Spectral Angle Mapper Algorithm]

Area of Science:

• Geoscience
• Remote Sensing
• Spectroscopy

Context:

• Spectral uncertainty in terrestrial objects introduces variations in feature spectra, affecting object recognition accuracy.
• The widely used Spectral Angle Mapper (SAM) algorithm, based on spectral curve similarity, does not account for spectral uncertainty, limiting its effectiveness.
• Accurate identification of terrestrial objects using hyperspectral data is crucial for various applications.

Purpose:

• To investigate the applicability of the Spectral Angle Mapper (SAM) algorithm in the presence of spectral uncertainty.
• To propose a modified SAM algorithm that addresses and overcomes the limitations imposed by spectral uncertainty.
• To enhance the accuracy and reliability of object recognition in hyperspectral remote sensing.

Summary:

• A modified Spectral Angle Mapper (SAM) algorithm is proposed to mitigate the impact of spectral uncertainty on object recognition.
• The modified SAM calculates spectral differences between test and reference spectra, considering the principle of minimum angle.
• Experiments using USGS mineral spectral data demonstrate that the modified SAM improves object recognition accuracy, particularly with local band combinations.

Impact:

• The modified SAM algorithm enhances the characterization and management of spectral uncertainty in hyperspectral data.
• It offers improved accuracy for terrestrial object recognition, especially in scenarios with significant spectral variations.
• The study validates the enhanced applicability of the modified SAM for diverse terrestrial objects and spectral conditions.