Abstract: In a solid-state image sensor in which a large number of pixel cells each comprised of a combination of a main photosensitive pixel having a relatively large area and a subsidiary photosensitive pixel having a relatively small area are arranged, if the subsidiary photosensitive pixel has a defect for any pixel cell, division photometry data during AE processing is read, and the defective pixel is replaced with a value obtained by dividing the output value of the main photosensitive pixel at the same position by a sensitivity ratio only for a section for which it is determined that the main photosensitive pixel is not saturated. Thus, the pixel value of a defective pixel can be accurately corrected without causing a reduction in resolution sensitivity compared to a conventional method of correcting a defective pixel using surrounding pixel information.