Researchers from Microsoft Research and the Italian Institute of Technology have introduced a method to turn a regular video camera into a depth camera. The researchers have developed a machine learning technique for estimating per-pixel depth using any conventional monocular 2-D camera, with minor hardware modifications. The approach targets close-range human capture and interaction where dense 3-D estimation is desired. This involves hybrid classification-regression forests to learn how to map from near infrared intensity images to absolute depth in real time. The software includes an algorithm that learns the correlation between pixel intensities and absolute depth measurements. The algorithm is implemented on conventional color or monochrome cameras. The hardware modifications required were removal of any near infrared (NIR) cut filter, and the addition of a bandpass filter and low-cost LEDs, both operating in a narrow NIR range.
Source: Microsoft Research