A reflected light image of an observed region and a fluorescent light image of the observed region based on fluorescent light that is generated from the observed region are acquired. The variation component of the acquired latest reflected light image is detected on the basis of the latest reflected light image and an old reflected light image. A correction process is performed on the luminance value of the latest reflected light image taking the variation component into account. A normalization process is performed on the luminance value of the fluorescent light image using the luminance value of the reflected light image after the correction process, which sharpens the image of the region in the observed region from which the fluorescent light is generated.