Abstract: A video signal having a group of frames of successive dependent frames and an independent frame following the dependent frames is coded per group of frames, by interframe and intraframe coding, and a data amount of the group is controlled by feedback and feed forward control. A quantization step number for the dependent frames is changed based on a total data amount of the group and a data amount of the dependent frames for the feedback control. A temporary quantization number of the independent frame is determined for the feed forward control. A data amount predictive value for each block in the independent frame is determined by an activity of an image in each block and the temporary number of quantization steps. The predictive value is subtracted from the total data amount. A subtraction result for each block is accumulated to output a positive compensation amount if the result is a positive amount and a negative compensation amount if the result is a negative amount.