An image recorded on a photo film is projected through an imaging lens onto an image sensor. In accordance with a selected mode, set values are read out from a data ROM and are applied to a band pass filter having programmable digital filters. Pixel signals from the image sensor are converted into evaluation signals that are correlative with spatial frequencies of the image, and are filtered through the band pass filter with the frequency band determined by the set values. The evaluation signals are integrated, and a peak value of integration values obtained at different focusing positions is determined to be an in-focus position.