| abstract |
A computer implemented image processing for normalising images is disclosed. The method comprises calculating, for each chunk of an image, a combination of the two largest contributions to the variations of colour of the chunk that approximates a representation of three additive colour component data elements of the each image chunk, thereby to provide a plurality of approximated image chunks. The approximated image chunks are used to determine (i) a first item of approximated image chunk data, vMin, that corresponds to a first one of the two largest contributions, and (ii) a second item of the approximated image chunk data, vMax, that corresponds to a second one of the two largest contributions, wherein the first item of approximated image chunk data and the second item of approximated image chunk data together provide a mapping transformation and for each image pixel, image pixel data is obtained comprising three additive colour component data elements of the each image pixel, and transformed using the mapping transformation to calculate normalised image pixel data for the each image pixel. |