Shape extraction system and 3-D (three dimension) information acquisition system using the same

A 3-D information acquisition method of using the images obtained by photographing an object from a plurality of viewpoints and information about the positions of said viewpoints to recognize the areas occupied by said object in the images, estimating a 3-D shape of said object on the basis of the areas occupied by said object, and acquiring 3-D information about said object, said 3-D information acquisition method comprising:

• carrying out a first recognition and estimation process of not only recognizing the area occupied by said object on the basis of the image photographed with a visual line from a first viewpoint among the images photographed from said plurality of viewpoints but also estimating a 3-D shape of said object;
• carrying out a second recognition and estimation process of not only recognizing the area occupied by said object on the basis of the image photographed with a visual line from a second viewpoint opposing said first viewpoint but also estimating the 3-D shape;
• and
• thereafter carrying out a third recognition and estimation process similar to said first and second recognition and estimation processes by using the image from said viewpoint which corresponds to said image not to be used in said recognition and estimation process,
• and which is sandwiched between visual lines producing the largest angle difference corresponding to the two images used in said recognition and estimation process, and which is closest to the angle being interpolated in the largest angle difference;
• and carrying out a fourth recognition and estimation process of repeating said third recognition and estimation process.

A 3-D information acquisition apparatus comprising:

• a photographing module which photographs an image of an object;
• a background lighting module which illuminates, directly or indirectly from behind said object, a range including at least all of the silhouette portion of said object or a part of the silhouette portion in the photographing range of said photographing module;
• a relative movement module which moves said object and said photographing module relatively and continuously in such a manner that said photographing module can photograph images of said object from a plurality of viewpoints;
• a relative position sensing module which senses the relative position of said object and said photographing module at oath of the viewpoints at which images of said object are photographed by said photographing module from a plurality of viewpoints;
• and
• a 3-D shape estimating module which recognizes an area occupied by said object in each of the images photographed by said photographing module from the plurality of viewpoints by using the images of said object photographed by said photographing module from the plurality of viewpoints and information about the relative position sensed by said relative position sensing module and which estimates a 3-D shape of said object by using the areas occupied by said object, wherein said background lighting module is turned on in photographing an image used to recognize the area occupied by said object,
• wherein said 3-D shape estimating module carries out a recognition and estimation process of not only recognizing the area occupied by said object on the basis of the image photographed with a visual line-from a first viewpoint among the images photographed by said photographing module from a plurality of viewpoints but also estimating a 3-D shape of said object, and then
• carries out a recognition and estimation process of not only recognizing the area occupied by said object on the basis of the image photographed with a visual line from a second viewpoint opposing said first viewpoint but also estimating the 3-D shape of said object, and thereafter repeats said recognition and estimation process using the image from said viewpoint which corresponds to said image not to be used in said recognition and estimation process,
• and which is sandwiched between visual lines producing the largest angle difference corresponding to the two images used in said recognition and estimation process,
• and which is closest to the angle being interpolated in the largest angle difference.