If a frame is coded as a frame-structured B picture or as two field-structured B pictures, we say that the frame is a B frame; it can be reconstructed from information in the current frame and the two previously coded I or P frames (i.e., the I or P frames that will appear before and after the B frame).