Either the complete motion vectors are encoded, or only the difference between the motion vector and the corresponding BL motion vector is encoded.

A motion vector predictor (44) is then determined from the at least one candidate motion vector predictor (44, 54).

At least one candidate motion vector predictor (44, 54) is provided for a current motion vector (34) to be encoded.

At least one candidate motion vector predictor (54, 64) is provided for a current motion vector (34) to be encoded.

A predicted vector generating unit (109) estimates the motion vector of a block to be encoded, thereby generating a predicted vector (125).

A variable length encoding unit (110) encodes the difference vector between a motion vector (117) and the predicted vector (125).

A block unit movement vector searching unit (4) detects a movement vector (BMV) by a block matching processing.

A movement vector correction unit (5) performs miniblock division search to generate a movement vector (BV) when a movement correction error exceeds a threshold value.

  The calculated prediction vector is added to a difference vector decoded from an encoded stream so as to calculate a motion vector.

Either motion vectors from base layer are reused, resulting in a scaling factor and an update vector, or new motion vectors are calculated.

One of the two predictive motion vectors is chosen as the predictive motion vector for current block.

A movement vector detection unit (104) detects a movement vector of a block to be encoded with respect to a first reference image.

The code amount associated with the motion vector can be suppressed.

A pixel unit movement vector generating unit (6) allocates a minimum differential component between current and previous frame signals to a pixel movement vector.

The pixels of the current field are shifted according to the motion vector and the pixels of the reference field are shifted in the reverse direction according to the motion vector.