這個問題在Quora上有過討論,也算是我本行接觸很多的技術。基本上,leap motion的主要原理是使用紅外LED+灰階camera的方式才采集數據,並生成3D數據。
這樣的技術並不算新鮮,但Leap的優勢是在軟件方面,當然由於使用了紅外LED+灰階camera,成本也會更便宜一些。由於只處理手部的3D,不像現在Kinect需要生成全身skeleton, 復雜的depth資訊,Leap的執行效率自然也會高很多,對處理圖形的DSP要求也不會特別高。
Kinect的技術是用激光+camera的方式實作3D建模。其實兩者的實作的總的方式是差不多的,無非是技術使用以及演算法上面的區別。如果Leap的技術做遠距離的,效能就不會有現在的那麽好了。
其實,現在有比Leap更好的技術,成本上比他們還要便宜,就是E-field。具體技術細節我就不羅嗦了,想學習的可以Google去。E-field在3D建模的效能上,XYZ座標的辨識速度上都很好,而且關鍵的是,這種技術對PCB沒有過多的空間需求,成本也更便宜。相比Leap,我更看好E-filed。
附上在Quora上的一些討論:
Hans Gerwitz, underperforming consumer
10 votes by
David Lojudice Sobrinho,
Mattias Petter Johansson,
Waldir Pimenta,
(more)Engadget's hands-on review included this:
We asked both Holz, and Buckwald about the underlying technology that enables such high-fidelity controls, and were told that it's an optical system that tracks your fingers with infrared LEDs and cameras in a way unlike any other motion control tech.
http://www.engadget.com/2012/05/...This is pretty consistent with the form factor of the prototypes and pre-production models that have been shown. Also, Holz (the CTO who has developed the magic math that reportedly enables the Leap) previously worked on a NASA project that used camera data to determine atmospheric composition. Determining the difference between air and finger seems pretty trivial compared to finding methane in CO2.