Emerging visual IoT solutions meet different application requirements

The Internet of Things (IoT) will involve a large number of edge devices. FOG nodes need to aggregate the processed data, and then send it to the cloud/data center, and then send the final instructions from the cloud/data center to the edge device via FOG. The proliferation of edge devices in the future and the addition of artificial intelligence (AI) for analysis and processing require lower power consumption and more intelligence.

The Internet of Things (IoT) will involve a large number of edge devices. FOG nodes need to aggregate the processed data, and then send it to the cloud/data center, and then send the final instructions from the cloud/data center to the edge device via FOG. The proliferation of edge devices in the future and the addition of artificial intelligence (AI) for analysis and processing require lower power consumption and more intelligence. Edge devices include smart homes, where a large number of emerging applications such as security, home appliances, assistants, and gesture recognition involve visual IoT, and intuitive/easy-to-use imaging subsystems are essential.

Image sensor leader ON semiconductor has unique advantages in the field of visual IoT, including a broad lineup of image sensors, industry-leading and continuously improving imaging technology, rich application expertise and field application engineering team support, etc., to meet the needs of different imaging applications , To promote the key progress of smarter, lower power consumption, and emerging visual IoT applications.

Imaging and machine vision requirements

Visual IoT covers the imaging seen by humans and the machine vision processed by algorithms. For example, security cameras show images that humans see, visual assistants belong to machine vision, and video doorbells provide imaging directly when the user is at home, and perform video recording and comparison when the user is out. Since the comparison is processed by algorithms, So it belongs to machine vision at this time.

For traditional imaging, a specific requirement is high image fidelity. For machine vision, the sex-to-noise ratio and sampling are important factors, requiring accurate capture of moving objects.

At the same time, imaging and machine vision also have some common requirements for image sensors: For example, under different lighting conditions, when strong ambient light and low light exist at the same time, good pixel performance is required to achieve an excellent user experience; usually a wide field of view is required , Sometimes fisheye correction is required; it is cost-effective.

Security is the largest application field of the future visual IoT

For security applications such as IP cameras, external cameras, and video doorbells, image quality, high dynamic range (HDR), low-light performance and near infrared (NIR) are key performance factors. Differentiating factors include low power consumption and imaging under daylight/starlight.

AR0237 for entry-level security applications

ON Semiconductor’s AR0237 is the most cost-effective image sensor for this type of application. It has 2.1 million pixels, a resolution of 1928 x 1088, can achieve 1080p/60 fps output, provides excellent low-light performance and HDR, and supports high Dynamic range of long and short interlaced exposures. After capturing the two images of long exposure and short exposure, they are combined by the image signal processor (ISP), and then the best exposure pixels of the two images are selected and combined into one image. Satisfactory image.

AR0237 has two versions, RGB and RGB-IR. Among them, the enhanced NIR sensitivity of the RGB-IR version enhances night vision capabilities, so that the camera does not need special color filters for day/night switching, and uses infrared difference spectrum technology to avoid color changes. distortion.

Super starlight AR0221: leading low-light sensitivity and signal-to-noise ratio

The 2-megapixel AR0221 uses 4.2 um back-illuminated (BSI) pixel technology to provide excellent 3-exposure line-interlaced HDR with a resolution of 1936 x 1096, support for a 1080p/30 fps frame rate, and a signal-to-noise ratio of up to 41.2 dB , Its 16:9 ratio, with bright colors and high contrast. Other features include: array frame selection, automatic black level calibration and on-board temperature sensor, industrial-grade operating temperature range, dual data interfaces in the form of 4-channel MIPI CSI-2 and HiSPi SLVS, and durable, reliable and robust iBGA Package.

Figure 1 illustrates why AR0221 can be called a super starlight product: both pictures were taken at night without moonlight and starlight, the illumination is less than 0.01 lux, the scene is the same and the same lens is used, and the AR0221 is used. The image of is obviously clearer than the image taken by AR0237, and AR0221 only uses 26 times gain, while AR0237 uses 67 times gain.

Figure 1: AR0221 and AR0237 imaging comparison

AR0430: With synchronous depth mapping function, low power consumption

AR0430 won the 2018 CES Embedded Technology Innovation Award, with a resolution of 2316 x 1746 and a linear full-well charge capacity exceeding 15Ke-, thus providing HDR, using Super Depth technology to achieve synchronous video and depth mapping, and being able to provide anything within one meter from the camera The 30 fps video stream and depth image of the object can realize the functions of controlling smart IoT devices with gesture analysis and establishing simple 3D models in augmented reality (AR)/virtual reality (VR).

The AR0430’s low energy consumption performance enables it to provide 4 million pixels and 30 fps output with only 125 mW; in the low-power monitoring mode, the power consumption is reduced to 8 mW. This is especially important in battery-powered applications.

In the standard imaging mode, AR0430 can present high-quality images regardless of day or night light conditions. The built-in function of the device enables users to obtain color images and depth images from a single image sensor at the same time, saving the need for another sensor that provides a separate depth mapping function, thereby reducing costs. The main application scenarios are 1080p 120fps slow motion video, 1080p zoom, video and HDR imaging.

AR0431: Integrated near infrared + (NIR+) capability, low power consumption

AR0431 is based on AR0430 with enhanced NIR+ sensitivity technology. As shown in Figure 2, the innovative process doubles the sensitivity of 850 nm and 940 nm light without affecting the sensitivity of visible light, thereby helping security and IoT camera designers to create a highly sensitive night mode. Enhanced NIR+ enables AR0431 to achieve better imaging even in the darkest light conditions. Compared with AR0430, AR0431 uses fewer NIR LEDs, which can save cost and power consumption, or run LEDs at a lower speed to save power. If you use the same LED as AR0430, you can get better details and capture more light.

Figure 2: Enhanced NIR+ improves sensitivity

AR0431 is based on 2.0 um BSI pixels and provides a variety of low-power modes with a frame rate of up to 120fps, which is very suitable for applications that require slow-motion video shooting. Its low power consumption makes it an ideal choice for battery-powered security cameras, sports cameras, in-vehicle driving recorders (DVR) and general-purpose surveillance cameras.

Home appliance imaging application: AR0330 achieves high-speed image capture

In the field of home appliances, common imaging applications such as refrigerators, coffee machines, and sweeping robots, key performance factors include the ability to support motion and HDR. Due to the demand for miniaturization, compact solutions should be adopted as much as possible to achieve differentiated competitive advantages. For example, ON Semiconductor’s small-size Full HD AR0330, with an optical format of only 1/3 inch, helps to achieve a compact design, with a resolution of 2304 x 1296, excellent color accuracy, low-light sensitivity and low noise level. High-speed image capture, and includes adjustable configuration (gain, frame rate and exposure), while maintaining low power consumption.

In addition, the aforementioned AR0237 and AR0430 used in the security field are also suitable for home appliance imaging applications.

“Assistant” is a very promising application in the visual IoT

For example, visual assistants, voice audio, service/care robots, etc. are all “assistant” applications. Quick focus and HDR are the key performance factors for such applications, and in-depth capability/intelligence is the key to achieving differentiation.

AR1335: Quick focus via PDAF

AR1335 adopts phase detection focus (PDAF) technology, which can quickly help focus the pictures in newspapers, magazines, and manuals and convert them into voice for users to hear when used in visual assistant applications. AR1335 supports 4K, 30 fps output, based on 1.1 um BSI pixels, power consumption in 13 million pixels, 30 fps mode is 270 mW, other features include: high sensitivity to achieve excellent low-light performance, high full well charge capacity ( LFW) Realize HDR, 3D synchronization control, support stereo video capture, etc. In addition to visual assistants, AR1335 can also be used in applications such as 4K video recording, body cameras, drones, and sports cameras.

Robot application

The key performance factor for image sensors used in applications such as entertainment, sweeping robots, and companion robots is visual positioning, that is, imaging of moving objects, and HDR.

The global shutter AR0144 solves the smear problem

As we all know, using a rolling shutter to expose a moving object will cause smear in the image. The global shutter can solve the smear problem. ON Semiconductor has a leading market position in the global shutter sensor market.
AR0144 is an ON Semiconductor’s global shutter image sensor that can be described as a star. It has been widely used in scanning, drones, AR/VR, machine vision, 360-degree cameras, depth cameras and other markets. It uses 3.0 um pixels to provide Leading global shutter energy efficiency, the power consumption is less than 250 mW when outputting 1 million pixels, 60 fps, it can capture accurate and fast images in static and dynamic scene applications, and the size is extremely compact.

As shown in Figure 3, the image of the business card in the rotating fan captured by the AR0141 with the rolling shutter has been significantly deformed, while the image captured by the AR0144 with the global shutter has no artifacts.

Figure 3: AR0141 vs. AR0144

Compared with competitors that also use global shutters, AR0144’s global shutter energy efficiency (GSE) is 5 times better than competitors, and its full well charge capacity (LFW) is 2 times better.

Figure 4: AR0144 provides best-in-class LFW and GSE

In dynamic applications such as drones, AR0144’s low power consumption can extend the charging time interval, and its imaging performance can better map the environment and avoid serious collisions. In the design of 3D cameras and AR/VR, AR0144 can provide a cost-effective and small-size solution for capturing dual images for 3D mapping or depth perception.

In addition, ON Semiconductor also offers two global shutter sensors, AR0134 and AR0135, in iBGA packages for applications with stricter reliability requirements. The key specifications of the three global shutter sensors are shown in the table below.

Table 1: ON Semiconductor’s global shutter sensor

Concluding remarks

Edge devices of the Internet of Things will tend to be smarter and reduce power consumption as much as possible in the future. A large number of emerging applications in the smart home market it covers will generate demand for imaging subsystems. Design challenges include uncontrolled lighting environments and the lowest possible power consumption. It will be crucial to form a strong ecosystem by combining sensors with system-on-chip (SoC), lenses, modules, and algorithms to facilitate development. ON Semiconductor’s advantage lies not only in its leading image sensor technology to solve challenges in light conditions, power consumption, etc., to provide differentiated solutions and to promote future trends, and to have a broad lineup to meet different needs, but also to provide development kits and Combine solutions from industry partners to help simplify design and reduce costs.

The Links:   MG30V1BS41 LQ9D001