In-depth analysis of anti-smog and fog-penetrating technology

    In 2013, "smog" became the keyword of the year. In January of that year, 4 severe smogs covered 30 provinces (autonomous regions and municipalities). In Beijing, there were only 5 days without smog. A report showed that among the 500 largest cities in China, less than 1% of them met the air quality standards recommended by the World Health Organization. At the same time, 7 of the 10 most polluted cities in the world are in China.

　　The haze is coming in force, and is generally more serious in cities, which poses a severe test to the video surveillance system. Its main impacts are as follows:

　　1. The reflected light from the surface of the object is attenuated due to scattering by atmospheric particles, causing the brightness of the object image to decrease;

　　2. The reflected light is forward scattered by atmospheric particles and participates in the imaging of other pixels, resulting in blurred images and reduced resolution;

　　3. Some atmospheric particles are large in size and become noise during the imaging process;

　　4. Natural light that is not related to imaging is scattered by atmospheric particles and enters the image sensor to participate in imaging. This backscattering effect causes image saturation and contrast to decrease and hue to shift.

　　With the increasingly severe smog weather in China, many surveillance cameras have lost their monitoring functions. Smog causes the monitoring distance and image clarity of the camera to decrease, and even causes the loss of key details of the monitored subjects such as vehicles and people. Once an incident occurs, it will be difficult to obtain effective information, which greatly reduces the practicality of the video surveillance system.

　　So how can video surveillance deal with the onset of smog? The answer is fog penetration technology. As the name suggests, fog penetration technology can still penetrate the smog interference in smog weather, allowing the camera to obtain a good monitoring effect without affecting the monitoring distance and image clarity. Generally speaking, fog penetration technology is divided into electronic fog penetration (also known as digital fog penetration) and optical fog penetration (also known as physical fog penetration).

　　Electronic fog penetration is a back-end image restoration technology designed based on the human visual perception model. It uses the ISP algorithm to repair the image and eliminate the fog to achieve a more transparent visual effect. Optical fog penetration uses near-infrared light for imaging. Since near-infrared light has a longer wavelength, it is less disturbed by atmospheric particles during transmission and can penetrate haze to present clear image details. In fact, these two fog penetration technologies are not new technologies developed to deal with haze weather. In the early days, in forests, ports, docks and other areas, due to frequent fogging, fog penetration technology was used to achieve long-distance clear monitoring. With the arrival of haze, fog penetration technology has gradually become a standard technology for high-end cameras. For example, all series of IPCs for outdoor applications of Uniview Technology support electronic fog penetration, and the highest-end HIC5600 series also supports optical fog penetration on this basis.

　　So what is the difference between electronic fog penetration and optical fog penetration, and what are the advantages and disadvantages of each?

　　Electronic fog penetration

　　Electronic fog penetration uses visible light imaging and uses the ISP algorithm to repair the image. Currently, ISP algorithms are mainly divided into two categories: image enhancement fog penetration technology and image restoration fog penetration technology. Image enhancement fog penetration technology does not consider the cause of image degradation and has a wide range of applications. It can effectively improve the contrast of foggy images and improve the visual effects of images, but it will produce more noise and even cause more image details to be lost. Image restoration fog penetration technology studies the physical process of foggy image degradation, establishes a degradation model, inverts the degradation process, and compensates for the distortion caused by the degradation process in order to obtain the optimal estimate of the original image, thereby improving image quality. This method is highly targeted and the defogging effect is natural. The key point and difficulty of the process is the estimation of the parameters in the model.

　　From the explanation of the electronic fog penetration algorithm, we can see that its advantages are: 1. The color of the picture is preserved; 2. No additional hardware cost is added.

　　However, the disadvantages of electronic fog penetration are also obvious. First of all, due to the weak penetration ability of visible light, it will cause large scattering when encountering water vapor and dust obstructions, resulting in partial information loss. At this time, even if the image is repaired and the fog is eliminated through the algorithm, the lost information cannot be restored "from scratch"; in addition, the electronic fog penetration has different fog penetration effects based on different algorithms. Simple algorithms have limitations in image restoration effects, while complex algorithms put forward higher requirements on hardware performance. Moreover, in terms of the overall level of current image restoration technology, electronic fog penetration can only be used in general foggy environments. Once the haze concentration increases, the effect of electronic fog penetration will be greatly reduced.

　　Optical fog penetration

　　Natural light is composed of light waves with different wavelengths. The visible range of human eyes is roughly 390nm-780nm, which is called visible light. The wavelength less than 390nm is called ultraviolet light, and the wavelength greater than 780nm is called infrared light. Light of different bands has different characteristics due to different wavelengths. Generally, infrared light has a longer wavelength and is less affected by atmospheric particles during transmission. It can penetrate a certain concentration of fog and smoke. The camera can use this part of light to achieve monitoring through haze. This is the principle of optical fog penetration.

　　From the principle, we can know that the advantage of optical fog penetration is that infrared light has strong penetration ability, which will not lose the details of the monitored subject and the image is clearer. But its disadvantages also exist: 1. Since infrared light has no corresponding visible light color map, the image presented on the monitor is black and white, so it is not suitable for places where color images are required; 2. The focus of optical fog penetration is to intercept near-infrared light imaging, and this part of the work is mostly done by the lens, and the fog-penetrating lens is expensive. The high cost greatly limits the large-scale application of optical fog penetration in actual projects.

　　New fog-penetrating technology

　　From the above analysis, we can see that electronic fog penetration and optical fog penetration have their own advantages and disadvantages, and complement each other. So is it possible to combine the advantages of these two fog penetration technologies to provide an economical and practical fog penetration camera?

　　The answer is yes.

　　Through a detailed study of the fog-penetrating performance of light of various bands, and the analysis of a large amount of test data collected under different concentrations of haze, and finally combined with the application requirements of monitoring fog-penetrating, Uniview Technology has newly developed the HIC5600 series of network cameras with built-in optical fog-penetrating and electronic fog-penetrating dual fog-penetrating functions. In close range or general foggy environments, this series of cameras eliminates the fog by turning on electronic fog-penetrating to obtain clear and transparent color images; once the haze has a serious impact, turning on optical fog-penetrating can still achieve a clear optical fog-penetrating effect even if only equipped with an ordinary lens, without losing any details. In addition, due to the built-in optical fog-penetrating component, the cost is greatly reduced compared to the traditional optical fog-penetrating lens solution, and it can be widely promoted and applied in industries such as safe cities. Due to the inconsistency of the focal planes of visible light and infrared light, this fog-penetrating camera also has a built-in automatic back focus (ABF) device, which is automatically triggered after the mode is switched to obtain the clearest image.

　　Future Trends

　　In actual applications, as an ISP image restoration technology, the fatal flaw of electronic fog penetration is that it cannot restore lost information when the haze is severe. This is determined by its imaging principle, and even the most advanced image restoration algorithm cannot improve it. Once the haze concentration in the environment changes, the effect of electronic fog penetration will be uncontrollable, which means that the environmental adaptability of electronic fog penetration is weak, and the cost-effectiveness of developing more advanced image restoration algorithms is not high. Therefore, the author believes that the electronic fog penetration technology based on monitoring needs will gradually converge.

　　Optical fog penetration has always been used in forest fire prevention, maritime affairs, and border defense. It has a long history of practical application and its effect has been recognized by users. However, it cannot be promoted on a large scale in other general industries due to its high cost. With the emergence of new technologies, the cost problem has been solved. It is believed that its application prospects in various industries will become increasingly greater.

　　In order to improve the optical fog penetration effect, in addition to further improving the image processing capabilities of the camera itself, the supporting imaging systems such as lenses and windows should also have a higher transmittance to infrared light. Therefore, the improvement of supporting systems and even further upgrading of sensors are the future directions of optical fog penetration.