Wednesday, March 26, 2014

Dual Aperture Announces its DSP SoC Partner

PR Newswire: Half a year ago, Dual Aperture has announced its partnership with Siliconfile (SK Hynix) to make a customized image sensor. Now the company announces its DSP partner - eWBM Co., to be designing a processing SoC to run Dual Aperture's image capture, refocusing, 3D image generation, depth estimation and gesture tracking algorithms.

Dual Aperture's image sensor technology is based on 4-color sensor design. The 4-color sensor is comprised of RGB and IR pixels. The sensor also uses separate apertures for RGB and IR pixels, thereby capturing data for two distinct images - one each in the RGB and the IR spectrums with varying sharpness, as shown in its patent application here.

The company's Vimeo video demos Dual Aperture's gesture control:

From the company's FAQ:

Q8. What are the advantages of DA approach vs. other gesture tracking devices like Kinect or Leap Motion?

Kinect and Leap Motion systems require multiple sensors with elaborate LED lighting sources. The Dual Aperture system requires only a single sensor using ambient light sources, which make it possible to implement gesture tracking in small form factor cameras, e.g. mobile device cameras. The DA system also works under direct, bright sunlight (unlike Kinect) and captures high quality color images (like conventional cameras).

Q12. Are there any cost advantages or disadvantages compared to other similar multi-utility cameras?

The Dual Aperture camera has lower manufacturing costs when compared to other similar depth measurement technologies that require additional sensors, special infrared light sources, and complex computing for image alignment.

Q16. Is the DA sensor available?

Yes. A 3.2M pixel mobile sensor sample is currently available. Other resolution sensors are currently under development and will be made available in 2014.

1 comment:

  1. The main problem is that 1) Gesture is not the killer ap for mobile 3D Imaging and 2) The killer applications for mobile that do need 3D (real world acquisition. See Tango) need more precision that can be obtained from single chip solutions like this and array cameras (the parallax is too small to get the long-range depth accuracy required).


All comments are moderated to avoid spam and personal attacks.