So sometime back I saw this video presentation of a new and, what I like to call it, novel method for extracting 3D structures from a single image. Part of the reason why this blows my mind, is that this approach is well defined for a specific scenario and it utilizes the best of both human brain and computer's processing power.
We have a great sense of depth perception of objects. Our brains are well trained to construct an object's three dimensional model, by just looking at pictures. This, however, is a trivial and a highly challenging task for computer algorithms. On the other hand, computers are capable of computing and interpolating data at a much faster rate than humans, given that the task is simple and fairly straightforward.
To achieve a robust approach for capturing a 3D model from a single 2D picture, this approach uses a well trained human brain to capture the broader idea of the object. This broader idea is captured using what is called the 3-Sweep method, where three sweeps give vital queues for computer to process. Following this, the energy of these queues are in some way minimized to fit along the major axis of the object. In this intelligent combination of two different procedures, a highly robust techique is realized.
The limitations for this approach are well defined and justified, which is another thing which I like and helps the author to achieve a highly accurate and robust method for the defined scenarios. This is exactly why I love my field and will continue my research as there are always possibilities for getting something unique and useful, all you need to do is think out of the box.
I am definitely looking forward to when this paper becomes available online.. will be a good read!