Electronic image processing with automatic color correction

Abstract

A colored original is scanned point-by-point and density values are generated for each point in the primary colors red, green and blue. The points are divided into groups and the density values for each primary color are averaged over the respective groups. The averaged density values for each group are transformed into a set of converted values including a luminance value as well as first and second chrominance values. The range spanned by the luminance values is divided into segments and each set is assigned to a respective segment. The sets are classified as color dominant or color nondominant by comparing the saturation value of each set with a threshold value. For each, segment the first and second chrominance values of the nondominant sets are respectively added to yield first and second sums. The first and second sums are shifted relative to the gray point to generate corrected first and second sums. Each pair of uncorrected first and second sums is converted into uncorrected red, green and blue density values whereas corrected density values are obtained from the corresponding pair of corrected first and second sums. For each primary color, the corresponding uncorrected and corrected density values define a point in a coordinate system of uncorrected density versus corrected density. By plotting the points from the various segments, a correction curve can be established for each primary color.