To easily and high-precisely detect the movement of a video signal.

A feature extraction part 3 obtains feature quantity by executing a timewise and spatial high pass filtering processing on the noticed picture elements of the signal s0 of a present field and a signal s1 delayed by one field. A neural network part 4 generates a signal showing movement from feature quantity and a movement judgement part 5 judges the presence of movement by comparing the signal with a threshold.