An image processing apparatus generates a normal image, when lighting correction of an image is performed using three-dimensional shape data. Three-dimensional shape data of a predetermined object is adjusted with a subject in position, based on a feature pattern detected in image data, and lighting processing for correcting a pixel value of the image data is performed based on the adjusted three-dimensional shape data, distance information of the subject, and a position of a virtual light source. When the three-dimensional shape data is adjusted with the image data in position, a region having a large irregularity in the three-dimensional shape data is preferentially adjusted in position.