Abstract: A method for the correction of the colors of an image, the image being split up into a plurality of image points, wherein a color to undergo maximum correction and a spatially limited domain around the extremity of the vector associated with it in the color space are selected, for each of the image points whose color is to be corrected a test is made to determine whether a color vector associated with the point has its extremity in the domain and correction by addition of a color correction vector is carried out for the color vectors whose extremity is located within the domain, the color correction vector resulting from a freely selectable vector and a superimposed quantity which makes the color correction vector maximum if the color vector to be corrected coincides with the vector of the color which is to undergo maximum correction and which makes it to be the closer to zero the closer the extremity of the color vector to be corrected is to the edge of the domain.