To avoid a code amount from considerably increasing or an extent of improvement of image quality from being reduced when applying conversion encoding to an image signal including a region form.

A dividing method determining means 102 inputs an image signal composed of M pieces of region and region information and determines a method for dividing the image signal into blocks with 1st-Nth sizes on the basis of the region information, and a dividing means 101 divides the image signal into the blocks on the basis of the determined dividing method. A conversion base correcting means 104 corrects a conversion base function that a conversion base holding means 105 holds, on the basis of the region information in the blocks, and a converting means 103 uses the corrected conversion base function to convert the pixel value of each block into a conversion coefficient. Thus, the number of coefficients to be transmitted can be remarkably reduced as compared with the case of using the conventional converting method for an arbitrary form region.