Method and device for detecting a color cast in an image

Method for detecting a color cast in an image, comprising the following steps:

• converting said image into a chromatic space of the type CIEL*a*b*;
• defining, inside said chromatic space, a region of the near neutral objects of the image converted into said chromatic space;
• by setting a given range of lightness values L and by calculating, in successive iterations, a given value of the radius of chroma in such a way that said region of the near neutral objects includes a percentage of the image greater than or equal to about 7%;
• determining the mean value and the standard deviation of the distribution, in a plane with L*=constant of said chromatic space, of the color 2-dimensional histogram of said region of the near neutral objects;
• analyzing the difference between said mean value and said standard deviation;
• detecting said color cast in the event that the difference between said mean value and said standard deviation is greater than a first determined positive value;
• characterized in that the method further comprises the steps of:
• performing a count of the number of distinct peaks of said histogram in the event that the difference between said mean value and said standard deviation is less than said first determined positive value and greater than a second determined negative value;
• and
• detecting said color cast in the event that the number of said peaks is equal to one.