At present, the five mainstream technologies of surveillance cameras are: wide dynamics, fog penetration, intelligent analysis, high pixel and starlight level. What is the overview of these technologies? Let's take a look at it with Xiaobian:
Wide dynamic
Digital wide dynamics does not achieve the goal of expanding the dynamic range of imaging in the true sense. Instead, the image post-processing algorithm of the software improves the contrast of the local area, which is generally implemented by the camera ISP module. The range of grayscales that we can discern with the naked eye is very limited. In fact, computers can distinguish very weak grayscale differences. Digital wide dynamics enhances these weak differences to the naked eye through image processing algorithms. After that, based on the CCD hardware technology, the double frame synthesis wide dynamic appeared. The solution is to use a CCD, but each point above is exposed twice in a single time, one long exposure (low shutter), one short exposure (high shutter) ).
Therefore, each point has two data outputs, called "dual output CCD". Using DSP's unique image processing algorithm, the appropriate portions of the two images are cut separately, and finally the superimposed composite is output and a dark and dark area is output. Clearly visible image. Whether it's digital wide dynamics or dual shutter width dynamics, its wide dynamics are not ideal.
With the evolution of DSP and CMOS technology, DPS uses a separate exposure and control technology for each pixel, combined with a multi-frame image captured by a CMOS sensor to synthesize a linear overlay of a complete image, compared to the two-exposure imaging of the CCD. A higher dynamic range. Numerically speaking, CMOS cameras using DPS technology have a dynamic range of 120 dB or even 140 dB. Wide dynamic technology has become an important indicator of the performance of a camera. For the time being, the standard wide dynamic function has become the consensus of all IPC vendors.
Through fog
With the deterioration of the domestic haze weather environment in recent years, the demand for fog-permeable cameras is very strong. Because optical fog-transparent lenses are expensive, in order to reduce the price of fog-permeable cameras, and to achieve better fog-passing effects, the mainstream IPC manufacturers have begun to study the camera video image fogging algorithm technology. The algorithm can realize the clear fog-free image by judging the concentration of smog by the gray degree of the local area according to the physical smog formation model. The algorithm allows the fog to retain the original color of the image, while greatly improving the image fogging effect.
Insight
HD network cameras have launched two or three intelligent analysis functions, such as motion detection and video occlusion, introduced in 2011. Today, almost all mainstream manufacturers' HD network cameras come standard with more than 10 intelligent functions. Of course, these intelligent functions are currently available. The vast majority of the standard is limited to mid- to high-end industry products. According to the market business application, these intelligent analysis functions can be divided into the following points:
1. Diagnostic intelligence analysis. The diagnostic intelligence analysis of HD network cameras is mainly for the common camera failures such as black screen, blur, pan/tilt control, picture freeze, video signal interference such as scene change, item legacy/disappearance, etc. for accurate analysis, judgment and alarm. . Diagnostic intelligence analysis technology is relatively simple to implement. Usually these intelligent functions are integrated in the front end. Of course, the back end, such as the NVR, also has diagnostic intelligence like this.
2. Identify intelligent analysis. The technology of HD network camera is biased towards the analysis and processing of static scenes. Through the core technologies of image recognition, image comparison and pattern matching, the extraction and analysis of related information of people, vehicles and objects are realized. In the application of identification and analysis of vehicles, it is mainly the license plate recognition technology. License plate recognition technology is widely used in parking lot entrances and exits, highway toll stations and other places. In recent years, it has developed rapidly: in conjunction with the traffic electronic bayonet system, the license plate recognition technology has been widely used for the capture of vehicle traffic violations, effectively reducing the number of vehicles. The number of vehicle traffic violations has greatly reduced the occurrence of traffic accidents.
3. Behavioral intelligence analysis. HD Network Camera This technology focuses on the analysis and processing of dynamic scenes. Typical functions include: vehicle retrograde, zone intrusion detection, personnel focus detection, squall line crossing detection, fast movement, personnel detection, and passenger flow statistics. Motion detection (VMD) is the "early intelligence" in this kind of intelligent analysis. VMD is judged according to the motion change of the pixel block in the video picture. Because it is a two-dimensional image intelligence analysis, the false positive is high and the mobile is not recognized. Whether the pixel block is interference or target, plus the difference in algorithm technology between security manufacturers, the accuracy of behavioral and cognitive intelligence analysis is generally not high.
High pixel
In 2010, the HD first year was only a 720P HD network camera. Until 2012, the mainstream security enterprise HD network camera was mainly 1.3 million and 2 million, and there were few 300 or more pixels. With the introduction of CMOS technology, HD network cameras have taken a shot of a strong heart. After 2013, 3 million, 4 million, 5 million, 6 million, 12 million pixel cameras have sprung up. This is technology. Innovation brings innovation in the product system. In the existing “Pearl River Deltaâ€, “Yangtze River Delta†and “Circle of the Bohai Sea†mainstream enterprises, almost high pixels have become the standard of all manufacturers, and even 4K has become the standard of all manufacturers. For the user, 4K is not just a visual experience and enjoyment, but the resolution of 4K is 4 times that of 1080P. If 4K camera and 1080P camera are used to shoot the same scene under the same field of view, 4K camera will use 4 The amount of information used by the 1080P camera is restored to the scene, and the picture is naturally clearer and closer to reality. From the perspective of "use", since the information volume of 4K pictures is four times that of 1080P, based on more information, more accurate intelligent analysis can be realized. Once 4K is deployed on a large scale, the accuracy of intelligent analysis can be A step up, and a richer and more amazing smart application will be realized.
Star level
The industry recognizes 0.001Lux and below as the star-level camera. The representative star-level camera is the TI DM8127/Ambar S2+ Sony IMX185 hardware solution, which is widely used in safe city, finance, hotel buildings, safe villages and ports. In projects such as highways, there is no need to install supplemental lighting facilities on a large scale, and you can get better nighttime HD color monitoring screens. Starlight-level illumination monitoring technology is mainly affected by factors such as lens, image sensor, and back-end image processing technology. All manufacturers of security enterprises are also upgraded from the following aspects:
1. Using a large aperture lens: The lens is an important part of the camera component. Its role in the low-light monitoring application technology is to focus the light of the target on the camera. The key to the low-light application and technology here is that the lens has a larger diameter. The larger the amount of light entering the lens, the larger the aperture of the lens can effectively increase the amount of light entering, thus giving the camera an ideal low illumination effect.
2, the choice of large target sensor: the essence of the camera is to convert light energy into electrical energy, and the core component of the quantification is the sensor, the role of the sensor is to photoelectrically convert the different intensity of light transmitted to it, and convert it into voltage information. Generate digital image information. The part of the sensor that receives light is naturally the core of the core. If the camera of the same resolution has a larger target surface area, the larger the amount of light per unit pixel, the stronger the noise suppression capability, and the better the image quality when shooting at low illumination.
3, good image processing technology: the previous camera uses the traditional 2D algorithm to achieve the noise reduction function, and now the 3D noise reduction technology, based on the original frame noise reduction, by comparing the images of the two frames before and after The filtering process is used to find the noise position and gain control, and the 3D digital noise reduction function can reduce the noise interference of the weak signal image. Since the appearance of image noise is random, the noise appearing in each frame of the image is not the same. 3D digital noise reduction automatically filters out non-overlapping information (ie, noise) by comparing adjacent frames. With 3D noise reduction camera, image noise will be significantly reduced, and the image will be more clear and clear, thus showing comparison. Pure and delicate picture.
Stitch Bonding Machine
Stitch bonding machine is a type of textile machinery used to create non-woven fabrics. It works by mechanically interlocking fibers together using a series of needles and threads. The machine can produce fabrics with a variety of textures and properties, such as thickness, elasticity, and strength. Stitch bonding machines are commonly used in the production of clothing, upholstery, and automotive textiles. They are also used in the medical industry to create wound dressings, bandages, and other medical textiles.
