A skin-tone image portion contained in an image is detected based upon the shape of the image of a human face. An average value of each of RGB values of pixels that constitute the skin-tone image portion detected is calculated. If the distance between a skin-tone-blackbody locus and a value that is the result of converting the RGB values obtained by multiplying the average value by prescribed coefficients is less than a prescribed value, these coefficients are adopted as coefficients for multiplying the RGB values of each pixel constituting the image. By using a value that is the result of converting, to a chromaticity value, the RGB values obtained by multiplying the RGB values of each of the pixels constituting the image by the prescribed coefficients, those pixels of the image that have values belonging to a zone in the vicinity of a point on a gray-blackbody locus that corresponds to a light-source color temperature estimated based upon the skin-tone-blackbody locus are treated as gray candidate pixels.