An image processing apparatus, which determines, for a combined image obtained by combining pixels of a given first image and pixels of an unknown second image either translucently or non-translucently using an unknown coefficient indicating a transparency, whether each of pixels included in the combined image is a translucently combined pixel, is provided. The image processing apparatus calculates, from pixel values of the combined image and the first image of respective pixels in a predetermined area including one pixel, pixel values of an image corresponding to the second image, calculates a total of differences between the calculated pixel values, identifies a coefficient used to obtain the combined image from the total of the difference, and determines that the one pixel is a translucently combined pixel when a value of the identified coefficient is larger than a predetermined value.