An image of a subject is taken by a solid image taking device including a first image taking element having a plurality of pixels and a second image taking element having a plurality of pixels smaller than those of the first image taking element. First and second evaluations on contrast obtained by integrating high-frequency components of first and second image signals representing an image of the subject taken by the first and second image taking elements are calculated. The focus is adjusted on the basis of the first and second evaluations on contrast so that the subject is imaged on the solid image taking device.