摘要 |
The present invention is a merge mode parameter derivation device for deriving a prediction parameter including at least a vector and a VSP mode flag, the device being characterized in that, when coordinates (xPb, yPb) and height of a prediction unit is assumed to be nPbH, the coordinates of a position A1 are derived as (xPb-1, yPb+nPbH-1), and, when a merge candidate N is A1 and a VSP mode flag for the position A1 is 1, the VSP mode flag is assumed to have effect. The merge mode parameter derivation device is also characterized in that, only when a luminance compensation flag is 0 and a residual prediction weight is 0, the VPS mode flag for a unit to be predicted is assumed to be 1. According to the present invention, an effect is achieved that as depth access in viewpoint synthesis prediction is reduced, a processing volume decreases. Furthermore, an effect is achieved that as suitable depth access is performed for each sub-block used in viewpoint synthesis prediction, coding efficiency increases. |