Motion compensating apparatus

A motion compensating method for performing motion-compensated prediction with fractional-pixel accuracy between pictures that constitute a moving picture, said method comprising:

• a high-order tap filtering step of generating pieces of pixel data with half-pixel accuracy, by successively performing filtering operations in a predetermined direction on pieces of reference pixel data to be transferred, the pieces of reference pixel data being included in each block;
• an intermediate output pixel storing step of storing, into an intermediate output pixel memory, the pieces of pixel data with half-pixel accuracy generated in said high-order tap filtering step;
• and
• a data transfer controlling step of controlling transfer of the pieces of reference pixel data,
• wherein in said high-order tap filtering step, pieces of pixel data with half-pixel accuracy is generated by successively performing, on pieces of pixel data with half-pixel accuracy inputted from said intermediate output pixel memory, filtering operations in a direction orthogonal to the predetermined direction,
• in said data transfer controlling step, the number of pixels in a horizontal direction is compared with the number of pixels in a vertical direction in the current block to be processed, and the pieces of reference pixel data used for the current block to be processed in the direction in which a smaller number of pixels are arranged are transferred to be used in said high-order tap filtering step,
• the current block to be processed is made up of horizontal M pixels×vertical N pixels, where M and N are positive integers, and
• in said data transfer controlling step:
• the horizontal M pixels are compared with the vertical N pixels;
• the pieces of reference pixel data used for the current block to be processed are transferred in the vertical direction to be used in said high-order tap filtering step, and then ones of the pieces of reference pixel data used for the current block to be processed are transferred in the horizontal direction to be used in said high-order tap filtering step, in the case where N is smaller than M;
• and
• the pieces of reference pixel data used for the current block to be processed are transferred in the horizontal direction to be used in said high-order tap filtering step, and then ones of the pieces of the reference pixel data used for the current block to be processed are transferred in the vertical direction to be used in said high-order tap filtering step, in the case where M is smaller than N.