摘要 |
To allow any image to be viewed naturally and to determine a scene structure as close to a real object as possible, frame memories store therein three types of basic depth models indicating the depth values of three basic types of scene structures, one basic depth model for each scene structure. A composition ratio is determined according to the high frequency component estimation values or activity values of the luminance signal of a received non-3D image that are received from a top activity unit and a bottom activity unit and, based on the composition ratio, the three types of basic depth models are composed. The R signal of the non-3D image is superimposed on the composed basic depth model in an adder to produce final depth estimation data.
|