The number of pixels in an identified pixel region is counted, a feature point of the pixel region is extracted and the number of the feature points is counted when the number of the pixels counted has been determined to be equal to or higher than a first threshold value, whether the counted number of the feature points is equal to or lower than a second threshold value is determined, features is calculated based on the feature point extracted from the pixel region when the number of the feature points has been determined to be above the second threshold value, and the first threshold value is changed when the number of the feature points has been determined to be equal to or lower than the second threshold value. Image similarity determination process can be stably performed without any degradation in determination accuracy.