Hyperspectral Neural Radiance Field Method Based on Reference Spectrum

The Neural Radiance Field (NeRF) method for datasets is gaining attention for its wide applications and research value. Hyperspectral images have also gained many applications in recent years. This paper proposes a hyperspectral NeRF method based on reference spectrum, combines both NeRF and hyperspectral imaging. The method fully utilizes the features of hyperspectral data to obtain more consistent spectral characteristics and higher-quality synthesized images. Various experiments demonstrate that our method effectively improves the quality and spectral consistency of images generated at new angles of views with various training set sizes, compared with the original hyperspectral NeRF method. Additionally, the present method provides a convenient way to render images under different light sources with various spectra, expanding the potential applications of hyperspectral NeRF.