Method for generating natural colour from false colour images based on normalized difference vegetation index clustering

Abstract

Spectral information such as colour is one of the main types of information used for the purpose of image interpretation. Multi-spectral images consisting of the red, green and blue channels can be combined to produce True Colour Composite (TCC) images. Colours in such images are very similar to the colours registered by human eye. Unlike TCC images, some systems have no blue channel and use other-type sensors such as Near-Infrared (NIR) or Short-wavelength infrared (SWIR). Combinations of such channels result in False Colour Composite (FCC) images. In optical images where one or more primary channels (red, green and blue) are missing, it is possible to approximate the absent channels by simulating it from the existing ones. Such images, known as Natural Colour Composite (NCC) images and are very similar to TCC images. In this paper we describe an algorithm for generating NCC images from multispectral images containing the red, green and NIR channels. The principal assumption is that the spectral-transformation coefficients are not unique, but depend on the Normalized Difference Vegetation Index (NDVI). Our experimental results indicate that this approach generates more natural colours when compared with other algorithms, confined by unique transformation coefficients. © 2013 American Scientific Publishers All rights reserved.