An image processing apparatus forming an image of image data on a recording medium, is characterized by, when forming the image by synthesizing the image data with specific information including an image different from the image data, varying the number of pieces of the specific information according to a size of a specific information imparting region to which the specific information is imparted. Thereby, when the specific information imparting region is small, such as in the case of a postcard and a name card, it is possible to surely detect the specific information by increasing the number of pieces of the specific information. Further, when reading an image, the image processing apparatus varies, on the basis of a detected number of pieces of the specific information, a threshold value for determining whether or not the specific information is adopted, according to a size of the recording medium. Thereby, even when the size of the recording medium is small and therefore the number of pieces of the specific information is small, the detected number of pieces of the specific information is made to exceed the threshold value by setting the threshold value to a low value, so as to enable the specific information to be recognized.