An image processing apparatus interpolates a signal of a missing color component based on first, second, and third color signals respectively generated by first, second, and third pixels of an image sensor, to generate color signals. These pixels are arranged in a matrix. The first pixels generate the first color signal of a first luminance component of white light. The second pixels generate the second color signal of a second luminance component of narrow band light. Density of the first pixels being higher than density of the second pixels and density of the third pixels. The apparatus extracts a specific frequency component signal from a color signal of the first luminance component among the color signals generated by interpolation, and adds the specific frequency component signal to a color signal of a color component different from the first luminance component depending on white light imaging or narrow band imaging.