Thermal detectors are getting small enough and inexpensive enough to consider being integrated into smartphones…
Thermal imaging and sensing technology will certainly be among the lines of defense against the Covid-19 virus, according to market research firm Yole Développement (Lyon, France), triggering a boom in the market for thermal technologies. Between 2019 and 2020, the thermal imager and thermal detector markets are projected to rise by 76 percent and 20 percent, respectively.
Automotive down, thermography and public surveillance up
The Covid-19 pandemic is reshuffling the cards. High-volume applications such as automotive and ruggedized smartphones, which were expected to boom, are showing signs of stagnation, especially as production shifts towards more cameras for fever detection applications. Based on industry indicators, Yole now predicts that thermal imagers will be a $7.6 billion market, up 76 percent year-over-year. Previous forecasts predicted a $4.5 billion market, up 8 percent year-over-year.
Thermal imagers — microbolometers — have been struggling to penetrate the automotive market, and the current crisis will worsen the situation. The first effect is mechanical, said Éric Mounier, fellow analyst at Yole. Due to the slowdown in car shipments, microbolometer shipments for cars will drop.
“Microbolometers for night vision have always been considered as a luxury option in cars and an expensive add-on, we believe that, in the current sanitary crisis, people will put less money to buy high-end cars.” The second reason is that “thermal imaging is not yet considered as an important ADAS system in cars. We talk about CMOS image sensors, radars, lidars, but night vision is still perceived as a redundancy sensor for ADAS.”
Sensors are key to unlocking autonomous vehicles. They, however, generate a ton of data, and systems are limited by the processing power. One solution is improving data quality. “With infrared cameras, you can see things that you could not see with a lidar or a vision camera. It adds more safety to the car at night or in bad conditions, but I think we still need to improve the resolution and the image database.”
Companies like FLIR are trying to push the adoption of thermal imaging in cars, but Mounier said, “we don’t see in the short term a strong demand for night vision in cars.”
The stagnation, or even decline, in the automotive market will be offset by the surging demand for surveillance and thermography systems linked to fever monitoring in airports, hospitals, public areas and warehouses. For thermal imagers, Yole said it expects that more than 1.5 million fever detection cameras will be deployed over the next four years. “There will be a peak this year due to the high demand for passenger screening in airports,” commented Dimitrios Damianos, Technology & Market Analyst at Yole. “The rest will be spread out in the next three to four years due to the various infrastructures adopting thermal imaging.”
The deployment will depend geographically on the privacy policies and practices and people’s perception country by country. “We believe that the move to infrared imagers will come quicker in Asia than in the US and in Europe,” said Mounier.
Not only will perception change over time, but two types of technologies can be used. One is a bispectral camera, which has a CMOS imager for normal color pictures and a thermal imager to detect people’s temperature.
“This poses a privacy problem because you take a picture of someone, you identify it with AI or some database and you associate fever with the person,” Damianos explained.
The second option is a single thermal imager, which does not reveal people’s information. “You can detect if a person has a higher temperature without needing to store his or her information.”
To perpetuate thermal imagers, some weaknesses need to be addressed. The temperature accuracy is one of them as thermal imagers detect temperature with a ±2 to 5°C accuracy. “Technically, in the thermal imaging industry, for applications such as thermography or public surveillance, we did not need very good accuracy,” said Damianos. But for fever detection, “you need to be around or below ±0.5°C, and ideally 0.1°C” to avoid letting feverish people through.
Yole also deplores a lack of AI for infrared imaging. “There will be a strong demand for AI in surveillance and, in the future, for transportation,” said Mounier.
Referring to a recent study by Intel examining whether AI can recognize people’s faces using thermal imaging, he said, “we are still in the first steps of using AI, because we don’t have enough images in stock.”
Besides, with thermal imagers, the resolution is too low for face detection. “Most of the demand today for thermal imaging is QVGA [Quarter Video Graphic Array]. We see people’s temperature, but we cannot track them to do face recognition.” For that, AI needs to be implemented. Today, Damianos added, “companies claim that they have AI, but we believe it’s marketing. A pure thermal AI is not there yet.”
As daily temperature checks have become the rule at school and company doors, the demand for thermopile-based thermometers and associated temperature guns have exploded. According to Yole, the thermal detectors market will grow by 20 percent, to $350 million in 2020. This compares with previous forecasts of 10-15 percent, to $320 million this year. “This extra growth comes from extra shipments of thermal detectors for thermometers,” said Damianos. The thermal detector market, which comprises thermopile and pyroelectric technologies, is usually stable.
The idea of integrating thermal imagers in a smartphone is not new. “We heard about that many years ago, and we know that, for instance, Huawei was in contact with a company doing microbolometers to see if there was any point in integrating thermal imagers in smartphones,” Mounier explained. In 2017, Apple came with a 3D capability, and “smartphone makers focused on the development of 3D sensors and forgot about thermal imagers in smartphones.”
Besides, there was no real use case to have a thermal imager in smartphones at the time. The first reason was educational, because “it’s very difficult for people who are not used to seeing thermal images to understand what they see.” People might panic or make assumptions. The second reason was the bulky size of the module and the prize of technology.
The technology is here, and the time is probably right, said Yole. Some thermal detectors or imagers such as FLIR’s Lepton are tiny enough to be integrated into a smartphone. And some smartphones such as the Caterpillar S60, co-developed with FLIR, already integrate a thermal imager, albeit for a more professional use case.
What is needed today is to improve the accuracy — better than 0.5°C — and to have reliable electronics and processing to avoid false alarm, along with maintaining the cost in the range of a few dollars. Cost would inevitably decrease with volume production, benefiting from large wafer scale manufacturing capabilities on 8-inch or 12-inch diameter wafers.
As in all crises, the Covid-19 pandemic will create winners and losers. According to Yole, the companies manufacturing the thermal imagers as well as the optics for the thermal cameras will see their shipments explode, as the demand skyrockets. And there is no risk of production or supply chain bottleneck since China is now up and running and has sufficient capacities to address the huge growth.
Asked if China could benefit from this “fever detection gold rush”, Mounier confirmed he has seen great progress in the microbolometer technology with first products developed and manufactured in-house. “It started a few years ago, and probably the current crisis will speed up the move,” said Mounier. There is indeed a willingness “to have in-house technology in China, and no longer be dependent on external companies like FLIR or Lynred because infrared is linked to security.” Looking one or two years ahead, Mounier said he would not be surprised if Chinese companies such as HIKVision and Iray Technologies had a significant share in infrared cameras.