PURPOSE: To improve the coding efficiency by dividing a picture into plural blocks depending on displacement of a vector between picture elements and calculating a zoom coefficient for each block so as to enhance the correlation between fields and frames to be coded.

CONSTITUTION: A motion vector detection section 101 detects a motion vector from a reference field #0 and a field #1. In parallel with the detection, field data #1 are written in a field memory 104. A block division control section 10 divides the block based on displacement of the vector between adjacent picture elements. A zoom coefficient calculation section 103 calculates a zoom coefficient for each block based on a motion vector of the fields #0, #1 detected by the detection section 101 and block division information from a control section 102. A magnification reduction processing section 105 generates the reduced picture of the field #1 for each area divided by the control section 102 to provide an output of the result to a coding section 106.