updated 01:30 am EDT, Fri March 30, 2012
Would use various techniques to gauge depth
A patent has been awarded to Apple for "Image Capture Using Three-Dimensional Reconstruction," a method that may improve upon existing 3D imaging devices such as still and video cameras by overcoming the weaknesses of the two main types of 3D sensors already in existence. As outlined in Apple's patent application, currently devices can approximate depth but can't accurately reproduce 3D geometry, while other sensors are better with shapes but bad with depth.
In short, says Apple, current sensors cannot differentiate between small objects positioned close to the lens or larger objects that are farther away. Apple's proposal for a solution involves a number of different embodiments, but essentially combines many of the previous ideas for determining depth or geometrical information into a single device or coordinated set of devices. Several of the described techniques involve post-processing of polarized and unpolarized images, along with a second or additional 3D sensors that would capture images from an offset angle giving post-processing programs more information to determine correct geometry and depth.
The camera may also employ additional depth-sensing technologies such as radar, lasers and Light Detection and Ranging sensors (LIDAR), and post-processing would utilize information such as color separation algorithms, stereo disparity maps, dichroic prisms, microlenses and of course multiple CCD sensors to increase the amount of geometic and depth information to amalgamate in post-processing. While none of the mentioned technologies are original, Apple's implementation relies on combining and coordinating implementations of previous technologies to enable more visual information gathering to happen and the processsing to occur much more rapidly than has been possible previously.
The technology has other implications. As a simple example, once the camera has detected a face or faces it could be signalled to take pictures through a facial gesture such as a wink, or delay capture until it detects that all faces are smiling with eyes open (ideas that are to some degree already in use in newer cameras). Three-dimensional models could be created by simply taking a few photographs of it from different angles or by rotating the object.
When combined with the higher-than-HD display quality that Apple and soon other manufacturers are employing, 3D images could become more important in future photography and display technologies compared to what is available presently. The patent names Brett Bilbrey, Michael Culbert, David Simon, Rich DeVaul, Mushtag Sarwar and David Gere as inventors, and was originally filed late in September of last year.