Illumination plunging, concept competition, AI racing… new variables in the security industry


Editor’s Note: This article is from the WeChat public account “Intelligent Things” (ID: zhidxcom), author Li Shuiqing, 36 氪 authorized to release.”Starlight”, “black light”, “AI ultra-light”, “Aurora”, these concepts in the security industry seem to be much more fired than the “low-light camera” itself.Regardless of the marketing name, in the final analysis, they use different methods to enable the camera to capture more white-like images at night.The low-illuminance camera can fill the deadly short board of the security night surveillance industry’s “night blind eye” and become one of the hottest technical breakthrough points in the security circle.According to statistics, nearly 70% of crimes occur at night. If many places such as roads, warehouses, and the wild cannot be monitored at night, the personal and financial security of cities and towns will only be in the shadow.The method of filling light has already existed, but the problems of “light pollution”, “energy waste” and “blind area” have been the pain points of night surveillance.Since 2014, the newly born “Starlight” has become synonymous with “low illumination” in the security field. Since 2016, low-light cameras have ushered in a stage of rapid development, and they cannot say that there is a big wind, but they are “new technologies” every year.The “Bright Spot Technology” gesture appeared in the most prominent position on the AMB booth.Under the chasing of security companies such as Hikvision, Dahua, Yushi, Tiandi Weiye, and Keda, the market demand for low-light surveillance cameras has been detonated.Many security companies and communication players have followed suit. They not only divide many fragmented markets, but also use artificial intelligence.

Dongfeng, showing a specific program.What are the magical features of low-light camera technology such as “Starlight” and “Black Light”?In the five years of steady development of low-light technology, how can Haida and other security giants and many security leaders stir up the market?In 2019, what is the low-light camera that is closely integrated with artificial intelligence and brought to the security industry?From “Starlight” to many variants of technology, to fill the shortcomings of the security eye For a long time, the “Security Eye” camera has a fatal short board of “night blind eyes”.According to statistics, nearly 70% of crimes occur at night. From 7:00 pm to 5:00 am the next day, it is a high incidence of crimes. The quality of camera nighttime imaging has become a shortcoming for the development of security big data.Starlight cameras came into being.In the usual sense, a star-level camera usually refers to a camera that can display a clear color image without any auxiliary light source in a low-light environment with almost starlight.The so-called star-level illumination is between about 0.01 lux and 0.001 lux, while the sunny, moonlight night illuminance is about 0.1 to 0.01 Lux.Similarly, “super starlight” and “black light” reduce the illuminance again.Generally speaking, when the illuminance is below 0.001Lux, it is “super starlight”, and when the illuminance is 0Lux, it is “black light”.
▲Compared with the effect of low-light camera and ordinary camera As early as 1998, the concept of low-light camera appeared. At the time, foreign brands such as Samsung and Sanyo were leading.In 2014, with the development of hardware such as photosensitive components, “Starlight” security surveillance camera products appeared on the market, and the protagonist at this time has become a local enterprise such as Hikvision, Dahua, Yushi, and Tiandi Weiye.They have released their own 2 million-pixel star-level camera products, and the camera’s “night blind eye” rehabilitation road officially opened.Simply put, the “night vision” of a star-level camera is optimized by multiple hardware dimensions such as lens, aperture, picture sensor and fill light.The lens and aperture limit the ability to transmit light and enter light respectively, while the image sensor passes the sensitizing ability, and the fill light can supplement the low illuminance of the camera by the fill light capability.▲The hardware dimension of the camera that affects its “night vision capability” (picture from the network) Although the camera “night vision capability” needs to be optimized from these dimensions, the hardware is not the most capable of stacking wood, soThe technical solutions and product lines of different security companies are different.For example, when the lens is selected, the larger the aperture, the better, but a balance between the amount of light entering and the depth of field.In addition, the use of sensors with large target surface, whether to use infrared and white light to fill light, select several lenses and sensors and other optical path design problems, raw data testing and ISP effect debugging are all factors that need to be considered by different manufacturers.It also directly affects the starlight camera effect.With the development of artificial intelligence technology, people have begun to use software methods such as machine learning algorithms to improve the night vision capability of the camera.The marketing concepts of “AI Ultra-light”, “Super Starlight” and “Aurora” prevailed in the market are different, and there are some differences in technical principles, but in general they are constantly optimized on the basis of “Starlight Camera”.At present, low-illumination cameras such as star-level cameras have been widely used in scenes requiring high picture definition in scenes with dark or no light, such as roads, warehouses, fields, farms, communities, factories, etc.For example, in the field of public security, the past cameras can only shoot black and white pictures, but the starlight cameras and even higher-level “black light” and “aurora” cameras can help the police through the color night vision effect to determine “what color clothes the suspect wears.””There are other issues, which greatly reduces the difficulty of solving the case.2016-2019: Illusion of illuminance, concept competition, AI racing from 2016 to 2019, with the update of components such as back-illuminated sensors, the star-level camera technology developed rapidly, and various manufacturers led by Hai Dayu began to launchVariant innovations such as Super Starlight, Black Light, and Aurora.In just a few years, security companies such as Haikang, Dahua, Yushi, Keda, and Tiandi Weiye have consistently pursued low illumination from the beginning, to improve camera clarity for intelligent identification applications, and to use AI to improve low illumination.The clarity and color quality of the camera actually experienced the evolution from “pursuit of image quality” to “pursuit of intelligence” under the urging of market applications.1. Hikvision: “Starlight” is eliminated, and the “black light” era is coming.
In 2014, Hikvision began the first shot of the “night vision” of security cameras.Hikvision has launched a star-level 3 million high frame rate gun-type network camera and a star-level 2 million pixel infrared network HD smart dome.In 2015, the star-level camera technology was applied to the Hawkeye series panoramic camera in the public security field.In addition, Hikvision has developed a “Starlight +” camera, paying more attention to the detection of moving objects, and increasing the signal-to-noise ratio by 30%.In 2016, Hikvision’s starlight camera has been widely used in national football competition security, museum monitoring, live broadcast of Qiantang River and other scenes. Its 4K star-level network camera won the special award for innovative products at the 2016 AMB.Just as the “Starlight” market was in a boom, Hikvision turned to the development and productization of “Black Light Camera” in 2016.At the Anbohui Hikvision booth, there was a small intricate darkroom with Hikvision’s “Black Light” series of cameras that could “see” the color of the objects in the 0Lux environment.▲In the same illumination environment, the comparison between the shooting effect of the starlight camera and the black light camera In March 2017, the Hikvision full color black light camera came as scheduled.It has a major breakthrough in technology. The dual sensor architecture using retinal imaging technology can mimic the retinal imaging of the human eye and solve the contradiction between color and brightness interference.At the same time, SpecEX dual-light fusion can get a picture that is less than the noise of the star camera.Today, Hikvision’s related product line has already derived products in the vertical field such as “black light explosion-proof camera”.At the same time, the black light camera has been widely used in the scenes of the typhoon “Litchma” flood control, 24-hour river channel monitoring, and monitoring of wildlife parks.Hikvision’s low-light camera product line experts said that the starlight camera has a glorious history, but with the appearance of black light cameras, the starlight camera will withdraw from the stage of the times.With the continuous improvement and optimization of low-light camera technology such as black light, market applications will continue to be normalized and scaled, and economical products will also use black light technology, which will gradually move to the low-end market and be popularized.
Then, Hikvision’s route to low-light cameras gradually focused on black light cameras, while starlight technology was mainly used as an auxiliary blessing for intelligent products.For example, its “AI partner” 4 megapixel star camera, star camera with internal pickup function and so on.2, Yushi Dahua: one side to follow, while pushing Chen as one of the security ace companies, Yushi Technology also launched a star-level camera in early 2014.
At the end of 2016, the high-definition star-light infrared dome network camera launched by Yushi not only achieves less-noise color imaging under extremely low illumination (0.005 lux), but also includes out-of-bounds detection, regional intrusion, entry area, exit area, flaw detection,Intelligent functions such as face detection and passenger flow statistics.In the face of the problem of difficult face capture in the current low-light environment, Yushi directly called Ban Haikang.Yushi believes that the “Double Sensor” architecture selected by a class of manufacturers achieves the technical and technical defects by improving the imaging effect after combining the two images into one.
The defect is mainly due to the color cast of the composite image, which cannot be applied to a place with no light, and the cost is high.In this regard, the solution given by Yushi is the self-developed “Extra-ISP” and “U-FACE” technologies.
“Extra-ISP” technology is based on the ISP of the new generation chip, adding time domain noise reduction algorithm and multi-dimensional sharpness lifting algorithm to reduce the smear of moving images in low-light environment, making people and other moving objectsThe details can be fully demonstrated.
Time domain noise reduction is the analysis of multi-frame images for noise reduction. The motion estimation is set to prevent the motion of the subject moving excessively.The “U-FACE” technology is supported by the new generation of deep intelligent core chips, and the new face exposure and face thumbnail enhancement algorithms are added to optimize the face in low-light environments and enhance the brightness of the face.Permeability and balance.This new ISP technology is applied to products such as MSI cameras created by Yushi.
In addition, Yushi seeks specificity from the storage of low-light cameras.It is understood that its 33 times star ball machine and 2 million super star camera use self-developed U-Code super coding technology, 2 million pixel camera only needs 1M code stream, is 1/4 of H.264, is H.2651/2, saving storage space without degrading image quality.Dahua’s new rhythm closely follows the pace of Haikang’s big brother.In September 2014, Dahua launched a star-level ultra-low illumination 2 megapixel HD camera.
In the following year, the 6-inch high-definition 2 million-pixel high-speed intelligent high-speed smart ball machine with 2 million pixels is said to integrate panoramic pan/tilt, tracking and positioning, intelligent analysis and other technologies. The budding “smart” can be applied in the construction of safe cities..In 2016, Dahua launched a differentiated 4K aurora camera that is almost synchronized with Haikang’s black light camera.The so-called “Aurora” is said to mean that the camera can achieve a more extreme night vision than the starlight.The Dahua 4K Aurora camera uses Sony’s 4/3-inch large target CMOS sensor. The large target surface is combined with the back-illuminated technology. Together with the ISP core processing algorithm, the minimum illumination reaches 0.0005Lux in color and 0.0001Lux in black and white.Dahua’s 4K Aurora Face Recognition Network Camera won the “Golden Tripod Award” in the year.The machine also displays face recognition and face attribute extraction based on Dahua deep learning technology related algorithms, supports structured information extraction of 8 attributes and 5 expressions, and can support up to 10,000 face database comparisons.In the next two years, “AI” became an important trend in the next few years.Based on artificial intelligence technology, “Hui” series AI intelligent star-level surveillance camera, dual-sensor Dahua 9-inch high-definition aurora network high-speed smart ball machine, suitable for community, hospital, bank and other commercial scenes of the wisdom series star-level face camera, etc.Wait, you can see the shadow of artificial intelligence.3, Hundreds of flowers, new technology and “ghost” coexisting components of the development of the head player low-light camera development conditions, the head players closer to promote the integration of low-light camera industry chain, with Phoenix Optics,The component manufacturers led by Yu Hao, Sony, APTINA, OV, and Sunny Optical have made great achievements in this process.Under the conditions of further development of component technology and further integration of the low-light camera industry chain, many security companies, communication companies and AI companies have also launched their own low-light camera solutions.Many security backbones are not limited to chasing the wind, but also actively seek their own specificity.For example, Kodak released the AI ​​ultra-light camera at the 2019 AMB.They believe that the mainstream intensified fill light, hardware stacking, and complex ISP processing on the market have different levels of light pollution, smearing, etc. Therefore, the self-developed deep learning image enhancement algorithm is adopted to directly input the sensor by learning a large number of scene captured photos.Data is used for image restoration.Based on AI ultra-light technology, Kodak has launched three new series of camera products.Including AI ultra-light light vehicle bayonet, AI ultra-light vehicle electric police and AI ultra-light light personnel bayonet.Kodak Technology’s low-light camera experts tell us that the AI ​​ultra-light low-light camera is different from the technical path of the star-light camera, and the application scene of the common star-light camera is slightly different. The AI ​​ultra-light camera is more suitable for the traffic field.Rather than monitoring public security scenes such as buckles.Some security companies are not only actively building their own low-light technology moat, but also working to integrate Starlight with other technologies to develop “Next Generation Starlight.”For example, Tiandi Weiye faces the single lens and single sensor to achieve the “double light path design”, which is to make the outline of the object clearer and the picture clearer. In the face of low-light tailing and frame loss, Tiandi Weiye uses self-developed TVP.The special action threshold setting of the technology, the linkage gain and the time domain airspace bilateral filtering and noise reduction to optimize the image quality; Tiandi Weiye is also considered to be the first player to mention the “Super Starlight” technology.In addition, Starlight + Car Identification, Starlight + Voice Prompt, Starlight + Alert, Starlight + Auto Dimming and other programs have greatly improved the ability of low-light cameras to cope with complex scenes.In 2019, the low-light camera “Hundred Flowers” will be seen at the AMB meeting. Not only will the planners go hand in hand, but the components such as Phoenix Optics and Yuyu in the upstream of the industrial chain will also have excellent locations, large exhibition areas and numerousLow-light components shine.Of course, under the trend, there are also all kinds of low-light black light and starlight cameras entering the market.For example, a brand of night vision camera issued by the owner shoots “ghost”.This also reflects the problem that the technical dimensions of low-light cameras are difficult to standardize, cost is high, and the market is mixed.
▲A brand of night vision camera shoots “ghost” complex scenes put forward new requirements, AI technology brings new power five years ago, let the camera “see” the color picture in the environment of reaching out and seeing the five-figure thingNow, not only can you see it, but you can also see it clearly, and even after you see it, you can identify it, and you can let the camera take the initiative to handle the task through alarms, voice prompts, and so on.From this point of view, as a market with a certain threshold of product direction, low-light cameras have developed very fast in recent years.Although the concepts of “Starlight” and “Black Light” have appeared in the Expo for years, there are new connotations and applications every year.Hikvision experts responsible for low-light camera products told Zhishen reporter that in the future, with the continuous improvement and optimization of low-light camera technology such as black light, market applications will continue to mature, toward normalized and large-scale applications, and economical.Products will also use black light technology, from high-end products to low-end, popular.
Cheng Yuewei, general manager of Tiandi Weiye Company, believes that Starlight technology has been widely used in smart cities, Xueliang projects, government agencies, prison institutions, border defense construction, park security, etc. because of its inherent advantages. In addition, Starlight is integrated with other technologies.To solve more system problems, this will be a trend.
Wu Yida, product manager of Smart City Division of Keda Technology, told us that low-light technology products are not what people think, but can solve many common problems.For example, in the field of traffic police, it can solve the problems of “light pollution disturbing the people” and “the road environment suddenly flickers the eyes of the driver”, so it is very practical.At present, relevant policies have also brought many opportunities for low-light cameras. For example, Shanghai has already promoted the application of low-light technology in the smart city landing, and believe that other cities will follow up quickly.At present, the landing of the starlight camera is mixed.On the one hand, starlight cameras have been widely used in many places where roads, factories, fish ponds, etc. are weak and require high-definition images. On the other hand, there are still some practical problems in the face of complex application scenarios.
For example, on the road, “vehicle headlights” is a difficult point for the landing of low-light cameras. Too fast and too strong light will bring challenges to the camera to see the environment and license plates.In addition, although the current black light camera, 4K star-level camera pixels have been able to support intelligent functions such as license plate recognition, but the resolution still has a lot of room for improvement in the existing hardware configuration, the manufacturers are increasingTechnologies such as sensors, updating ISP technology, and developing new fill lights to optimize imaging results.In addition, due to the current level of hardware technology, the cost of true star-level and black-light cameras is still relatively high, and the equipment with intelligent functions is more costly for customers.Although in the big scenes such as smart transportation and safe city, such as the cancellation of the highway toll booth, many scenes match the low-light cameras, but customers will still choose more cost-effective considering the actual cost.Program.The security AI trend has brought new impetus to the development of low-light cameras, and also brought new players to the game.From this year, Keda brought “AI ultra-light” products and Shangtang to “face recognition small black house”. In the AMB meeting, “AI kicks out”, AI technology also brought a fresh weather to the security industry..In this kind of weather, non-big players are getting more and more out of the technical path different from the giants, and then form their own intelligent low-light camera technology moat.While the opportunities for next-generation converged technology applications such as “Starlight+Intelligent Recognition”, “Starlight+Intelligent Dimming”, “Starlight+Warning” are coming, although the security giant still has an advantage in terms of industrial ecology and channel accumulation,Other security companies are also expected to get a big cup in the security fragmentation market. As for the actual application effect, we will wait and see.Wang Ning, general manager of Tiandi Weiye Industry, said that with the development of AI intelligence, the application of low-light cameras will also see tremendous growth.On the one hand, software applications such as AI intelligent recognition require low-light camera technology of “hard-core soft assistant” to support, on the other hand, AI software technology will also promote low-light cameras to detect their own limit performance again.In addition, Wang Ning believes that the system scheme of power rate suppression and specific application scenarios is also the two major trends in the development of low illumination cameras.Conclusion: Low-light camera technology into industrial technology game point In recent years, “Starlight-class” low-light camera has changed from concept to standard products of various manufacturers, and black light, aurora and other more extreme low-light technology has become the focus of attention of various manufacturers..From pursuing a lower illumination environment to pursuing a better-looking picture, and pursuing the ultimate clarity and color reproduction to help AI intelligent recognition applications, the development of low-light cameras has been leaping forward.In the field of low illumination surveillance cameras, traditional security companies have a first-mover advantage.Low illumination monitoring technology involves many technical dimensions, and has certain industry thresholds. Traditional security players with early development and sufficient industry chain resources and channels lead the trend.With the increasing application of AI in the security industry, non-giant security companies use AI algorithm technology to open up new low-light surveillance camera technology paths, and can also show differentiated play, thus dividing the fragmented security market.The ever-developing low-light surveillance camera technology has enabled the security industry to continue to extend the boundaries of the scene. The 8-hour camera has become a “24” hour working system. Many wilderness fields, factory fish ponds, and deep lane corners that were previously not well monitored are not.In addition to the extraterritorial land, problems such as “light pollution” and “energy waste” are also solved.Although there are many complex scenarios that need to be considered in the application of technology, this fragmented market situation also presents different opportunities for players with many different technical routes.