An image processing apparatus includes an image acquisition unit that acquires image information representing a fluorescence observation image of a specimen stained with hematoxylin-eosin, a spectrum generation unit that generates a plurality of spectra each representing a wavelength distribution of fluorescence intensity in a plurality of pixels in the fluorescence observation image, a pixel extraction unit that extracts at least two pixel groups with a feature of a particular spectrum from the plurality of pixels, and an image generation unit that generates an image based on the extracted pixel groups.