In a video encoding method according to the present invention, an encoding mode controller 102 determines an encoding mode of each image on the occasion of encoding a moving picture consisting of a plurality of images, and a region partitioning unit 104 determines a region structural unit for partitioning the image into multiple regions, based on the encoding mode. A region determining unit 116 defines the regions on the basis of the region structural unit and information thereof is encoded. Then an encoder 105 compression-encodes pixel data included in the regions, in encoding units to generate compression-encoded data according to the encoding mode, and outputs the resultant data.