Provided is a technique that makes it possible to extract the image of a document area from a read image with better accuracy. An image processing apparatus capable of achieving this includes a reading unit for reading a document, which has been placed on a contact glass, as read image data; a first area determination unit for deciding a provisional document area by subjecting the read image data to detection processing for detecting a document area included in the read image data; and a second area determination unit for deciding a fixed document area by generating binary image data in an area corresponding to the provisional document area of the read image data, and executing detection processing for detecting a document area, which is included in the read image data, based upon the binary image data.