Abstract: Input image data is divided into a plurality of blocks. An orthogonal transform device subjects the input image data to orthogonal transform every block. A memory temporarily stores output data from the orthogonal transform device. A first quantizer quantizes output data from the memory. A coding device codes output data from the first quantizer into variable-length code words. A code amount accumulation device accumulates a code amount of output data from the coding device. A second quantizer separates the output data from the orthogonal transform device into K different groups, and assigns M different scale factors to the respective blocks and quantizes the output data from the orthogonal transform device in the blocks with the related scale factors, where K and M denote predetermined natural numbers. A block code amount calculation device calculates a code amount per block.