From this year's perspective, the intelligent security system in the era of the Internet of Things is more versatile and efficient. Cross-border applications of video surveillance, access control, and alarm have also become the norm. From Internet companies to security companies, all have launched AI open platforms, and the importance of ecosystem construction has become more clear. In addition, the overall industry's thinking and layout has undergone a new change. Comprehensive data collection and processing performance is the constant pursuit. While focusing on solutions, the ability of individual products has also become the focus of the exhibition. In general, product technology iteration and application innovation have a holistic summary this year. Multi-camera, low-light full-color has opened a new chapter in product form and application.
Multi-camera, the future development trend of front-end equipment
Before the artificial intelligence landed in the security field, from the product form of the camera, the common models can basically be divided into gun machine, dome camera, dome camera, gimbal, panorama and other major types. Functionally speaking, it is basically qualified to check the video clearly after the event. However, with the continuous development of AI technology, intelligent event detection and even trend analysis and early warning let users see the feasibility of preventing incidents. Now, although the user's usage scenario has not changed, the demand for intelligent services in the same scenario is constantly increasing. For example, the gun-ball all-in-one class breaks through the scene limitations of traditional monocular cameras, integrates panoramic perspective and detail close-up data, and combines large scene monitoring and detail linkage snapshot applications. In addition, due to the inconsistencies in the characteristics such as the moving speed of different targets in the same scene, different shutter parameters and fill light conditions may be required. Therefore, simple guns and domes can no longer meet the needs of all-round intelligent monitoring in the same scene, and multi-cameras have emerged.
Multi-camera cameras actually have prototypes, such as multi-eye fisheye, binocular head, bowl-shaped eagle-eye, multi-eye fusion panorama and other cameras. The Expo was first proposed. Multi-camera is a software and hardware technology innovation driven by changes in user needs, which meets multiple intelligent needs in the same scenario.
When Hikvision proposed a multi-camera system, and simultaneously launched a variety of intelligent products such as global, dual-cabin, and gun-ball integrated machines, many other companies in the industry also launched similar products, such as Huawei released a series of compound eye shapes. Cameras, Dahua ’s Hubble Watcher Panoramic Camera, Tiandiweiye ’s 4K Starlight Intelligent Binocular Global Ball, Yushi Technology ’s Four-eye Panoramic Camera, and Zhongxing ’s Gunball Linkage All-in-One. Each shows its magic.
From intelligent IoT data collection to value data entering the information network, cameras that take into account different perspectives, different parameters, and different functional requirements in the same scene are bound to be the development trend of future data collection front-ends. Multi-camera can take into account the multiple intelligence requirements in the same scene. Scene definition and multi-camera integration are inevitable forms of future AI cameras.
Low-light full-color, necessary conditions for intelligent realization 24/7
In the data era, cameras should have the "four full" capabilities of all weather, full scene perception, full factor extraction, and full data correlation. Regardless of rain, snow, fog, or day and night, to adapt to all kinds of complex light, the camera is indispensable for high-quality imaging capabilities under low light. From this year's perspective, the low-light full-color technology of various giants has initially reached the level where face recognition can be achieved.
In the traditional low-light full-color implementation, the first is the use of LED fill light, the light is dazzling, it is easy to affect the sight of people entering and leaving, and will also cause the vigilants of the crime. The second is the use of infrared fill light to obtain black and white images without color information. This year's Expo, head companies have launched low-light or "0" pollution to achieve full-color imaging. Hikvision has applied for a patent for mild light supplementation this year, making black light technology even more powerful. Dual-light fusion and gentle fill-light dual-pronged approach, farewell to light pollution and clearer color targets, enabling full structured information extraction under low light; Huawei's intelligent color fusion technology based on AI algorithm, combined with intelligent exposure technology and invisible infrared fill light, Realize full-color clear snapping effect without eye perception and true “0” light pollution. Based on Kodak's ISP image modulation technology, Kodak adopts a self-developed deep learning image enhancement algorithm. At the same time, it uses super-large target surface image sensor technology and unique fill light technology to effectively avoid light pollution and finally achieve night ultra-high-definition monitoring and intelligent functions. Tiandi Albert fused AI images of black and white and color pictures through a unique intelligent algorithm. The most realistic picture can be restored, and the true full-color effect can be achieved without relying on the infrared light to fill the light. Yushi uses infrared and white light dual light fill light, infrared light to collect brightness, white light to collect color, can output full-color fusion image without white light flash at night.
Regardless of which technology is used, such as black light, low light, invisible infrared fill light, dual light fill light, and the same goal, they have basically achieved full-color image collection under ultra-low illumination. The collected license plates and faces meet the identification comparison. The pixels also initially achieved intelligent early warning capabilities in the dark.
AIoT provides the foundation for data fusion
With the deepening application of AI technology and the development of big data and cloud computing technology, security video surveillance has gradually transformed into the video Internet of Things. From video structuring to the most multi-dimensional data perception proposed this year, in addition to traditional video surveillance, data sources have added Wi-Fi, RFID, temperature and humidity sensing, access control, anti-theft alarm and other dimensions of IoT information. For the data application of the government public security industry, data resources can be called from various departments within the public security according to different permissions. The advent of the 5G era has also provided transmission assistance for the collection of mobile video surveillance and other IoT information. The rapid development of 5G technology, sensor technology, radio frequency identification technology, and intelligent embedding technology has made it possible to connect everything, and the world where everything is connected has allowed the sublimation ability to be sublimated. AIoT multi-source data convergence perception provides a strong foundation for data fusion. Adding to the construction of the digital twin city.
Application of 5G in security
This year, 5G has been mentioned a lot in security, but it is well known that 5G network deployment has just begun in China, so there are almost no products in the real application industry. It is undeniable that the development of 5G technology will definitely have an impact on the security industry, but whether it is imagined remains to be discussed.
At the beginning of setting, 5G technology was defined to carry out research for characteristics such as ultra-high speed, high reliability, and low latency. Similar to the traditional security industry, which has been constrained by low-bandwidth and low-rate transmission systems, the era of intelligent security is still trapped For network transmission systems. But can the application of 5G technology really change the status of data transmission in the security industry?
Video surveillance has high requirements for the bandwidth of data transmission. In actual projects, if the cost of the wired network deployment is much higher than the cost of the 5G network, there is no doubt, but it is too early to talk about the cost of a 5G network that has not yet been deployed. . For access control systems, wired deployment is basically required in actual projects, so the demand for 5G networks can be ignored for the time being. As for the anti-theft alarm system, the amount of data is small, and the stability and reliability of data transmission are high, so the application requirements for 5G are not urgent. In view of various factors, the application of 5G in security is still immature at this stage, and it is not clear how the specific development trend will be in the future. Perhaps with the completion of the 5G national network deployment, there will be greater advantages in terms of transmission speed, network stability, unit cost, etc., and innovation will come only when it can meet the demands put forward by intelligent security development.
Voiceprint recognition, rising star
At present, the global biometrics industry is huge, and the market size of voiceprint recognition alone is close to 10 billion US dollars. It is expected to exceed 20 billion US dollars in 2020, accounting for 22.4% of the entire biometrics market. As a non-contact biological collection feature, voiceprint is one of the fastest-growing technologies following the face and fingerprint recognition technology. Voiceprint recognition has been frequently mentioned this year. With the development of pickup technology and speech recognition technology, the Ministry of Public Security began to promote voiceprint technology to the whole country. Similar to fingerprint database and DNA database, the construction of voiceprint database is also focused.
Voiceprint recognition has strong technical expertise and a wide range of applications. Currently, voiceprint recognition companies include micro-nano sensing, Alibaba Express, and sound technology. Micro-nano sensing has launched a super far-field pickup solution that combines far-field pickup and video In combination with monitoring, we have cooperated with many mainstream security companies to launch audio-video linkage products and solutions to achieve audio-visual linkage based on abnormal sound triggers.
Access control AI technology mature
The application of artificial intelligence in access control is deepened. In addition to the application of artificial intelligence algorithms in the software platform, the more obvious performance is the intelligent application of front-end equipment. Chip-level AI empowerment and the blessing of algorithms have made the application of face recognition in access control more popular, and the market size has grown rapidly.
Access control solutions that incorporate AI authentication technologies such as face recognition, QR codes, fingerprints, and NFC have become the norm. The application of 3D face recognition technology is also becoming more and more widespread. Since 3D face recognition is based on the collection of 3D stereo image features, the recognition information is more abundant. Therefore, compared with binocular living recognition, the accuracy is higher and the Improved. In the future, the deep integration of services based on three-dimensional stereo software interfaces will also be one of the trends for future integration of access control systems.
Burglar Alarm Fusion Video Application
In recent years, the development speed of the anti-theft alarm industry technology has been slowing down. This year, artificial intelligence began to land in the alarm industry. Alarm products broke the boundary with video surveillance. The combination of monitoring and detectors was applied more closely. Similar visual smoke detection appeared The terminal monitors the detection equipment.
Last year, manufacturers successively introduced product solutions that focus on low power consumption, wide coverage, large capacity, and low cost, and mainly apply wireless technologies such as NB-IoT and LoRA. This year, the development of 5G technology supports the connection of more intelligent devices and a larger range of real-time verification of alarm intrusions. But in a & s interview, manufacturers questioned the stability of wireless transmission. 5G can further accelerate the speed of data transmission between systems and improve the stability of device connection. It is especially important for security applications with higher cost of delayed signals. However, whether the stability and reliability and data security can be guaranteed, alarm manufacturers are still waiting to see. .
In addition to wireless transmission, with the continuous progress of artificial intelligence and big data technology, burglar alarms have also made great progress in data analysis applications. After the alarm platform operates through artificial intelligence algorithms, more sensors can be connected or more devices can be controlled. The alarm is no longer simply an alarm function. More diversified application scenarios, such as home furnishing and intelligent monitoring, provide users with diversified product services.
Anti-theft alarm, as one of the four core systems of security, plays a pivotal role. In the era of intelligent security, anti-theft alarms have also begun to be combined with artificial intelligence and deeply integrated with video surveillance and access control to promote the intelligent upgrade of the entire security industry.