Strategy for pictorial digital image processing

DETAILED DESCRIPTION I claim: 1.
A method of pictorial digital image processing comprising the steps of: providing an original image; gathering preliminary information; constructing one or more of an image-dependent flare model, an image-dependent non-linearity model, an image-dependent preferred reproduction model, and an output-dependent preferred reproduction model in accordance with the preliminary information, wherein an image-dependent non-linearity model is constructed by the steps of: constructing a scaled image based on the original image; deriving image statistics from the scaled image, wherein the step of deriving image statistics from the scaled image comprises the sub-steps of: transforming the scaled image using preliminary information about a capture device used to provide the original image to produce scaled image focal plane data; determining significant statistical values of the scaled image focal plane data; determining flare characteristics of the original image; creating input linearization LUTs using the preliminary information and the flare characteristics of the original image; transforming the scaled image into linear scene integrated channel radiance data using the input linearization LUTs; normalizing the linear scene integrated channel radiance data using the preliminary information to produce normalized channel radiance data; and transforming the normalized channel radiance data into a standard color space; and processing the original image in accordance with the preliminary information and the one or more of an image-dependent flare model, an image-dependent non-linearity model, an image-dependent preferred reproduction model, and an output-dependent preferred reproduction model to produce a processed image from the original image, wherein the step of processing the original image further includes taking into account the image statistics derived from the scaled image.
2.
The method of claim 1 further comprising the step of creating a luminance channel by combining all channels of the normalized channel radiance data