Qualcomm and Qijing Optoelectronics Release 3D Depth Sensing Solution

This is really an age of looking at the face. It used to be just a face. Now the machine is starting to look at the face. Recently, "Face Recognition" has caused widespread discussion on social media, and the various sections of the road have not been idle, and each has its own magical powers, setting off another climax.

Although Duanzi is just a joke, from the endless stream of segments, we can also see that there are still many doubts about the security of “Face Recognition”.

However, there is no doubt that the era of "brushing the face" has come. From the "unmanned supermarkets" that were quite hot not long ago, to the automatic vending machines that "pay for the face", "face recognition" has been widely used in many fields. And we can already predict that a large wave of new machines will be equipped with a "face recognition" system. So, how is “face recognition” realized? How does the machine do "seeing the face and recognizing people" like a human being?

The realization of "face recognition" technology is first of all due to the major breakthroughs in deep learning in the field of "computer vision" , and "face recognition" is an important research topic of computer vision. With hundreds of millions of image big data generated by the smartphone era and deep learning algorithms, computers can now achieve a face recognition rate of more than 99%, which is very close to the recognition rate of human eyes.

Thanks to the breakthrough of "computer vision" technology, we have been able to apply "face recognition" to smartphones. However, to achieve high-precision and experienced face recognition, a better solution is to use "high-resolution, low-power active 3D depth sensing camera system." By projecting high-density infrared structured light, it can draw a 3D image of a human face, and then compare the image with the database to confirm the identity of the user. Thanks to the use of infrared structured light, this solution enables fast and accurate face recognition even in dimly lit environments. With the machine learning algorithm, even if the face changes its appearance due to time and environment, the mobile phone can recognize you.

Thanks to Qualcomm's accumulation of computational photography and computer vision, as early as mid-August of this year, Qualcomm has pioneered the development of deep sensor camera technology for the Android ecosystem . Announced the expansion of the Qualcomm Spectra module project and the next-generation Qualcomm Spectra ISP that supports computer vision, augmented reality and computational photography. The new ISP and camera modules combine to deliver outstanding image quality and new computer vision use cases that take advantage of deep learning techniques and background blur.

In the following August, Qualcomm and Qijing Optoelectronics jointly released a high-resolution 3D depth sensing solution. It combines Qualcomm Spectra's technology and expertise in computer vision architecture and algorithms, as well as Qijing Opto's complementary technology in wafer optics, sensing, drive and module integration capabilities, providing fully integrated SLiMTM (Structured Light) Module) 3D solution. SLiMTM is a one-stop, complete 3D camera module that delivers high-resolution, high-accuracy real-time depth sensing and 3D point cloud generation in both indoor and outdoor environments.

The second-generation Qualcomm Spectra ISP series and the new Qualcomm Spectra camera module are expected to be integrated into the next flagship Snapdragon mobile platform. The SLiMTM 3D camera is expected to be in volume production in the first quarter of 2018.