An image processing apparatus executes color balance correction of an image composed of multiple pixels. The image processing apparatus comprises a representative value calculation unit and a correction processor. The representative value calculation unit calculates, based on signal values of the pixels included in a skin tone area that includes pixels indicating human skin within the image, a representative value representing hue and chroma of the pixels included in the skin tone area. The correction processor executes color balance correction of the image using execution correction amount set based on normal correction amount obtained by multiplying a difference between the representative value and an ideal value designated as a value representing ideal hue and chroma of human skin by a reduction coefficient. The value of the reduction coefficient is set such that the value of the reduction coefficient can vary in response to the representative value.