The application of artificial intelligence has entered the "deep water area", and a large number of computer vision technologies have also "come to the forefront". The realization of computer vision is inseparable from the support of image sensors, in this regard, global shutter sensors have become the preferred imaging technology with competitive costs and accurate measurements.
Computer vision will move towards the global shutter
Computer vision, as a branch of artificial intelligence, enables computers and systems to extract meaningful information from digital images, videos, and other visual input data, and to act on or make recommendations based on that information.
Lin Guozhi, technical marketing manager of ST Asia Pacific Imaging Division, introduced that from the perspective of computer vision application scenarios, there are mainly four categories, one is depth sensing, which is used for application scenarios such as face recognition and 3D scanning. The second is biometrics, an application that collects information from various parts of the human body and uses it for identity recognition, including faces, palm prints, and irises. The second is AR/VR/MR, which further heats up under the concept of metaverse. Fourth, robots and industrial controls, such as drones and cleaning robots, need to avoid collisions during use, and require object detection and scene analysis capabilities, including barcode recognition.
The implementation of the above applications will be achieved through a key carrier, which is the global shutter sensor.
The reason why the global shutter is "on the way", Lin Guozhi explained that as one of the two technologies of the sensor, the roller shutter sensor is designed to capture still images and video shooting, so it has a very high resolution and color processing capabilities. But its disadvantage is that it is taking images line by line, shooting and exposure time is too long, if the subject is a fast moving object such as a car, the image may be distorted, and the power consumption is too high, not suitable for computer vision. The principle of global shutter is completely different, it is to shoot the entire image at once, all pixels are exposed at the same time. Moreover, the imaging effect is accurate, the exposure time is short, and the power consumption is also low, and the future computer vision will move towards the global shutter.
In response to this market demand, ST has launched two types of products: one is VD55G0, with a pixel of 400,000, and the other is VD56G3, with a pixel of 1.5 million. Lin Guozhi added that these two types of products have three major advantages: first, the sensor 940nm has the highest quantum efficiency, which leads to extremely high NIR sensitivity; second, the resolution of the square sensor is optimally matched to the lens; and finally, ST's sensor has a short shooting time, reducing system power consumption while ensuring accurate imaging.
It should be noted that the roller shutter pixels are commonly below 1.1 microns, while the global shutter is usually above 2.2 microns. Therefore, Zhang Chengyi, senior manager of the technology market of ST Asia Pacific Imaging Division, analyzed that in the traditional consumer photography and imaging fields, roller shutter sensors should still be very popular. Moreover, automotive ADAS or rearview mirrors will still be dominated by roller shutter sensors, but global shutter sensors will be widely used in areas such as driver and passenger monitoring, automotive entertainment information systems, and automotive safety.
Comprehensive coverage of global shutter applications
Based on the advantages of global shutter sensors, they have found new places of use in different fields such as consumer, industrial, and automotive.
But obviously, different industries have different needs for image sensors and need to divide and conquer. Zhang Chengyi pointed out that the application of smart home needs to be considered carefully, because users do not want cameras to monitor around the clock, but they need the comfort brought by sensors. In response to this, ST's global shutter sensor has developed a series of functions for human-computer interaction, such as people counting, face recognition, etc.
In the field of notebooks and PCs, Zhang Chengyi believes that the significance of applying global shutter technology is that it can carry out facial recognition, gesture control, eye tracking, comfort monitoring and privacy protection. In addition, ST offers ToF and ALS products for presence detection, battery life, automatic screen brightness adjustment and privacy protection.
It has to be said that the demand for cameras in smart cars is developing rapidly. According to Yole, the market penetration rate of in-car cameras will be about 10% at the end of 2021, and will reach 50% by 2024.
In the field of in-vehicle cameras, Euro NCAP, a vehicle evaluation standard system, divides in-vehicle surveillance systems into two categories: one is the driver monitoring system DMS, and the other is the cabin/occupancy monitoring system CMS/OMS. Cabin monitoring + driver monitoring is a combination of CMS, OMS and DMS, supporting RGB and NIR with a single camera.
In response, ST has introduced two generations of global shutter sensors for automotive applications.
According to reports, the first generation includes four types of products: VD5661A has a resolution of 1.6 million, mainly used for DMS. The VD5761A has a resolution of 2.3 million, as belonging to Mono can also be used for DMS. In addition, there is the VD6763A with color RGB and the VD1762A with RGB-NIR integration. Zhang Chengyi said that these products have three major advantages, one is high contrast, and being able to output images with extremely high contrast means being able to obtain more accurate information. Second, the data management of RGB and NIR is very flexible. The third is high dynamics, so that clear images can be obtained in bright and dark environments.
The second generation of VB56G4A has a resolution of 1.5 million, mainly used for driver monitoring, has three major advantages: first of all, high sensitivity, NIR's high sensitivity means that there is no need for a strong LED light source, because LED energy consumption will bring higher power consumption; secondly, the sensor area is smaller, which means that the camera module can also be made smaller and easier to install inside various mechanical devices; and finally, the integration of intelligent automatic exposure algorithms, without manual operation.
"In the future, ST will continue to reduce the pixel size, optimize power consumption, and continue to innovate and integrate at the algorithm and process levels to further expand the market." Zhang Chengyi finally stressed.
It can be said that the ST global shutter sensor has its proprietary pixel technology, higher square resolution and a variety of embedded computer vision functions to fully meet the requirements of computer vision applications. Moreover, ST provides not only global shutter technology, but also optical technologies such as flight time and ambient light sensors, and customers can integrate these different products into various scene applications according to their own needs to adapt to future application needs.
(Proofreader/Li Yan)