Depth estimation data generating apparatus, depth estimation data generating method, and depth estimation data generating program, and pseudo three-dimensional image generating apparatus, pseudo three-dimensional image generating method, and pseudo three-dimensional image generating program

The depth estimation data generating apparatus comprising:

• a generating unit which generates images of a plurality of basic depth models respectively indicating depth values of a plurality of basic scene structures;
• a composition ratio calculating unit which calculates a composition ratio between the images of the plurality of basic depth models by using statistics of at least some pixel values in a predetermined region in a screen of a non-3D image, so as to estimate a scene structure of the non-3D image whose depth information is neither included explicitly nor included implicitly like a stereo image;
• a composing unit which composes the images of the plurality of basic depth models generated by the generating unit according to the composition ratio and generates a basic depth model-composed image;
• a high frequency component detecting unit to which an object signal indicating object information about concavity and convexity of a pseudo 3D image is supplied from the non-3D image, and which detects a high frequency component indicating an edge of the object signal;
• a gain generating unit which generates a gain having a smaller value as a value of the high frequency component detected by the high frequency component detecting unit is higher;
• a compensated object signal generating unit which multiplies the object signal and the gain generated by the gain generating unit and generates a compensated object signal wherein a level of the object signal is compensated for;
• an adding unit which adds the compensated object signal to the basic depth model-composed image composed by the composing unit and generates depth estimation data for generating an image having a viewpoint different from the non-3D image, the image constituting a pseudo 3D image with the non-3D image;
• a high frequency component comparing unit which compares and determines which threshold value division range from among a plurality of threshold value division ranges defined by threshold values includes the value of the high frequency component supplied from the high frequency component detecting unit for each specific period in which the object signal is pre-set, and outputs a flag indicating a threshold value division range having the highest threshold value including the value of the high frequency component in the each specific period;
• and
• a gain calculating unit which calculates a ratio corresponding to the threshold value division range indicated by the flag supplied from the high frequency component comparing unit as the gain, wherein ratios corresponding to the plurality of threshold value division ranges are pre-set, each ratio having a smaller value as a threshold value corresponding to each ratio is higher.

A depth estimation data generating method performed by a depth estimation data generating apparatus, the method comprising:

• calculating a composition ratio between images of a plurality of basic depth models respectively indicating depth values of a plurality of basic scene structures, by using statistics of at least some pixel values in a predetermined region in a screen of a non-3D image, so as to estimate a scene structure of the non-3D image to which depth information is neither included explicitly nor included implicitly like a stereo image;
• generating a basic depth model-composed image by composing the images of the plurality of basic depth models according to the composition ratio;
• detecting a high frequency component indicating an absolute value of a difference between values of predetermined pixels in an object signal, the object signal indicating object information about concavity and convexity of a pseudo 3D image, and the object signal being supplied from the non-3D image;
• generating a gain having a smaller value as a value of the high frequency component detected via the detecting of the high frequency component is higher;
• generating a compensated object signal wherein a level of the object signal is compensated for, by multiplying the object signal and the gain generated via the generating of the gain;
• and
• adding the compensated object signal to the basic depth model-composed image composed via the generating of the basic depth model-composed image to generate depth estimation data for generating an image having a different viewpoint from the non-3D image, the image constituting a pseudo 3D image with the non-3D image.

A non-transitory computer readable storage medium having recorded thereon a depth estimation data generating program which is realized by a computer, the program comprising:

• calculating a composition ratio between images of a plurality of basic depth models respectively indicating depth values of a plurality of basic scene structures, by using statistics of at least some pixel values in a predetermined region in a screen of a non-3D image, so as to estimate a scene structure of the non-3D image to which depth information is neither included explicitly nor included implicitly like a stereo image;
• generating a basic depth model-composed image by composing the images of the plurality of basic depth models according to the composition ratio;
• detecting a high frequency component indicating an absolute value of a difference between values of predetermined pixels in an object signal, the object signal indicating object information about concavity and convexity of a pseudo 3D image, and the object signal being supplied from the non-3D image;
• generating a gain having a smaller value as a value of the high frequency component detected via the detecting of the high frequency component is higher;
• generating a compensated object signal wherein a level of the object signal is compensated for, by multiplying the object signal and the gain generated via the generating of the gain;
• and
• adding the compensated object signal to the basic depth model-composed image composed via the generating of the basic depth model-composed image to generate depth estimation data for generating an image having a different viewpoint from the non-3D image, the image constituting a pseudo 3D image with the non-3D image.