High-definition fog machine The effect of fog on video surveillance

The effect of fog on video surveillance It is generally believed that the atmospheric medium is mainly composed of air molecules, water vapor and aerosols. Aerosols are dispersed systems of small particles suspended in a gas. Certain particles are highly hygroscopic and act as condensation centers for water vapor, the size of which is related to ambient relative humidity, water vapor supply, and the level of condensation that occurs from collisions. Due to the difference in the size, type and concentration level of various particles in the atmosphere, various weathers such as Clearness, Haze, Fog, Cloud and Rain are generated. Under sunny weather conditions, the light reflected from the surface of the object will not be affected by various components in the atmosphere to cause scattering, absorption, reflection and other phenomena, and can directly reach the imaging device to obtain clear and fog-free images. In foggy weather, light reflected from the surface of an object is affected by particles suspended in the air on its way to the imaging device. Aerosol particles are the main element of haze and the basic reason for the degradation of image quality. Its influence mainly includes the following aspects: (1) Aerosol particles have a scattering effect on light, and the scattering loss attenuates the intensity of "transmitted light", resulting in a drop in the contrast of the image. (2) Due to the non-uniformity of aerosol particles, spherical waves are distorted into aspherical waves, resulting in blurred images and reduced edges and details. (3) The particle size of aerosol particles is large, and the image of the particles itself cannot be ignored, and can be approximately understood as "noise". (4) Due to the effect of repeated scattering, the scattering of the aerosol particles to the imaging light will be superimposed with the original forward scattering, resulting in a certain ambiguity. 2. Comparison and analysis of real-time video fog penetration technology and other fog penetration technologies The currently known fog removal algorithms can be roughly divided into two categories: one is a non-model image enhancement method, which achieves the purpose of clarity by enhancing the contrast of the image to meet the requirements of objective vision; the other is model-based. The image restoration method, which examines the cause of image degradation, models the degradation process, and adopts se processing to finally solve the image restoration problem. Video surveillance system fog penetration technology formula The current enhanced methods to achieve the typical methods of fog treatment include: histogram balancing, filter transformation methods and methods based on fuzzy logic. The histogram balancing method, in which the globalization method has a small amount of computation but insufficient enhancement of details; some balancing methods are effective, but may introduce problems such as block effect, large amount of calculation, amplified noise and difficult control of algorithm effects. The filtering and transforming fog-penetrating algorithms can obtain relatively good processing results after partial processing, but they have a huge amount of calculation, consume a lot of resources, and are not suitable for devices with high real-time requirements. The method based on fuzzy logic does not work well enough to penetrate the fog. Enhancement-based methods can improve image contrast to a certain level and improve legibility by enhancing areas of interest. But this method fails to compensate for the reason of the image degradation process, so it can only improve the visual effect and cannot achieve a good fog-penetrating effect. HD fog-proof movement At present, the methods based on image restoration mainly include the following categories: filtering method, maximum entropy method and image degradation function estimation method. The filtering method, such as the Kalman filtering method, has a large amount of calculation on the whole. The maximum entropy method can achieve higher resolution, but it is nonlinear, computationally expensive, and difficult to solve numerically. Most of the image degradation function estimation methods are designed based on certain physical models (such as the atmospheric scattering model and the haze model of polarization characteristics). , and the final solution obtains the resulting image in a fog-free state. This limits the application of such methods in real-time monitoring. Security products have been used in various complex scenarios and severe weather. All-weather real-time monitoring has put forward more stringent requirements on product portability, power consumption, disposal effect, and self-compliance of disposal. A good video defogging technology should integrate the technical advantages of image enhancement and image restoration on the basis of the atmospheric transmission model, so that ideal image effects can be obtained and used in practical engineering.