Can the “spare tire” of 3D facial recognition move into the mainstream?

The 3D facial recognition market is expanding rapidly. According to a forecast from Trend Force, the 3D sensing market will grow exponentially, reaching $10.89 billion in 2020 and $18 billion by 2023. The penetration rate of 3D facial recognition in the smartphone market is also expected to increase from 2.1% in 2017 to 28.6% in 2020.

The technology of choice for facial recognition

Although the pie is undoubtedly sweet, whether you’re able to “have your cake and eat it too” depends on whether or not that you have the technical cutting edge. Currently, there are two main technological options for 3D facial recognition in smartphones – one is a structured light solution, and the other is a time-of-flight (ToF) solution.

Of the two options, the most market-ready and commercially viable is the structured light solution. The widespread commercial application of structured light technology can be largely attributed to Apple, who took the lead in abandoning fingerprint authentication and adopting structured light 3D facial recognition technology for its iPhone X. Market feedback went from initial curiosity and doubt to the current total acceptance. Clearly, this technology can meet the criteria of picky users. Since Apple's move was presumably made after trialing many options, it should be a safe bet for other manufacturers. Thus, structured light 3D facial recognition already has many takers.

On the other hand, the ToF solution is regarded as a "spare tire." Nobody pays much attention to it. However, that can be partly attributed to its latecomer status. Latecomers have to work much harder than newcomers at proving themselves. Hence for the past two years, ToF has been trying very hard to prove itself.

Figure 1. Schematic diagram of 3D face recognition scheme for ToF mobile phone (Image source: Internet)

In terms of its working principle, 3D structured light technology uses a near-infrared laser to project light with certain structural features onto the object to be photographed, and then collects it in a special infrared camera. Since the light of the structural features changes in accordance with the depth of the object, depth information can be obtained by comparing and calculating the acquired phase of the image, thereby obtaining the 3D structure of the measured object.

When ToF technology is used for 3D facial recognition, the surface source continuously emits light pulses. When light is reflected upon encountering an impermeable object, the light source can be measured by measuring the time the reflection takes to reach the receiver, thereby enabling the distance from the object to be calculated and obtaining a 3D image of the object that has “depth.”

In theory, structured light technology is superior in accuracy and resolution, but ToF has many advantages that cannot be denied. For example:

• Relatively speaking, structured light technology requires more post-processing after obtaining information about the image to calculate the depth while ToF measurement is more direct and requires no post-processing. With ToF, the delay is shorter and the response is faster.
   
• The ToF solution, which does not require post-processing, also saves costs and power consumption associated with the processor, significantly reducing the technical threshold.
   
• The ToF solution requires fewer components and is smaller, making it more popular in mobile applications with space constraints and also more practical for mechanical devices.
   
• ToF is less susceptible to ambient light interference than optical solutions such as structured light.

After comparing the characteristics of the two technologies, some people have concluded that structured light technology is more advanced and more suitable for static use situations, making it the first choice for facial recognition in front-facing cameras on smartphones. On the other hand, since the ToF solution has lower noise at long distances, it is thought to be more suitable for dynamic use situations. Thus, ToF is deemed more suitable for rear camera lens placement on mobile phones, where it can show its skills in AR and VR applications.

For ToF technology, AR and VR are still far into the future, while front-facing camera facial recognition is already an existing market waiting to be tapped. In the 3D facial recognition market, clashes between structured light and ToF technology are inevitable.

ToF technology is already here

Today, ToF technology suppliers are fully prepared to tap the growing market and win over future users. For example, Infineon's REAL3™ series of 3D image sensors is an advanced solution that is ready for commercial applications.

Given that it has ToF's technical advantages, REAL3™ offers many more features of its own. For example:

• The REAL3™ series supports a minimum of 38,000 depth points, each of which has a measurement depth of millimeters and an error rate of less than one in a million, making it safer and more reliable.
• For resisting ambient light interference, REAL3™ uses patented SBI (Suppression of Background Illumination) technology to instantly eliminate all irrelevant signals for excellent camera performance.
• The camera module of the RREAL3™ series only has two main components – the REAL3™ image sensor and lighting circuit. No mechanical base line between the two is required, thereby greatly simplifying assembly and debugging while enhancing mechanical stability and mechanical strength.
• Naturally, this type of photographic lens module is also smaller, only 11mm × 7mm × 4mm in size, thus saving valuable smartphone space.
• In addition, REAL3™ has an integrated eye-safety circuit that meets safety requirements and reduces the BOM, unlike other solutions that require external eye-safety circuits.

Figure 3. Infineon's REAL3™ series 3D image sensors make the entire ToF camera module simpler and more robust (Image source: Infineon)

REAL3™, based on ToF technology, has already met with success in commercial applications. The front camera module of the LG G8 ThinQ mobile phone uses the REAL3™ solution. Infineon also revealed that the next-generation REAL3™ ToF image sensor will use advanced microlens technology to increase sensitivity and further optimize size and power consumption.

Of course, compared with structured light technology, the latecomer ToF still has a long way to go in terms of technical fine-tuning, commercial verification and supply chain support. However, well-known analyst Ming-Chi Kuo predicts that Apple is expected to adopt ToF technology in an iPhone in 2019. If this rumor becomes a reality, it will be a major boon for ToF's commercial expansion. Imagine, the “spare tire” of 3D facial recognition turns mainstream – now that would make for a very interesting scenario.