Abstract

Multispectral images record detailed color spectra at each image pixel. To display a multispectral image on conventional output devices, a chromaticity mapping function is needed to map the spectral vector of each pixel to the displayable three dimensional color space. In this paper, we present an interactive method for locally adjusting the chromaticity mapping of a multispectral image. The user specifies edits to the chromaticity mapping via a sparse set of strokes at selected image locations and wavelengths, then our method automatically propagates the edits to the rest of the multispectral image. The key idea of our approach is to factorize the multispectral image into a component that indicates spatial coherence between different pixels, and one that describes spectral coherence between different wavelengths. Based on this factorized representation, a two-step algorithm is developed to efficiently propagate the edits in the spatial and spectral domains separately. The presented method provides photographers with efficient control over color appearance and scene details in a manner not possible with conventional color image editing. We demonstrate the use of interactive chromaticity mapping in the applications of color stylization to emulate the appearance of photographic films, enhancement of image details, and manipulation of different light transport effects.