System and fast response access control method based on defogging artificial intelligence (AI) image analysis

Abstract

A system and a fast response access control method based on defogging artificial intelligence (AI) image analysis are provided. The invention realizes the scene mode judgment of the current image through the AI learning technology, an image local area division strategy adapted to scene mode is adopted, local scene depth differences are reduced, the local area division strategy with optimal adaptation to the image scene minimizes the error caused by the assumed local area scene transmissivity equalization, significantly reducing the blocking effect of the defogging image, and the fine adjustment algorithms such as soft matting and guidance filtering are not needed, which saves the computational resources of the access control system and accelerates the access control response speed. At thesame time, according to the invention, whether the defogging algorithm is invalid in the current image based on the scene mode can also be judged, and the parameters extracted from the fogless image under the same scene mode are referenced to carry out the modification of the relevant parameters in the defogging algorithm, thereby improving the robustness of the access control system.

Description

technical field [0001] The invention belongs to the technical field of artificial intelligence, and in particular relates to a defogging-based AI image analysis system and a quick-response access control method. Background technique [0002] When collecting outdoor images in foggy weather, the light absorption and scattering by fine particles in the air will reduce the contrast of the image, shift the color tone, and blur the picture content, which will inevitably bring difficulties to the rapid identification and analysis of the image content. come to obstacles. Especially for the access control system of smart communities and smart buildings, because it mainly extracts the fine eigenvalues ​​of the surface texture of the user's facial image to realize identity authentication, in haze weather, the above eigenvalues ​​are greatly disturbed, and it is easy to Misrecognition, delayed recognition, or no recognition occurs. In particular, the frequency of haze weather in some ...

AI Technical Summary

Problems solved by technology

[0014] However, the existing image defogging technology based on the dark channel principle still has defects. First, for the scene transmittance map t(x,y) in the formula, it is converted into the scene transmittance t( Δ), that is, it is assumed that the scene transmittance of each pixel in each local area is the same. As described above, the scene projective rate is determined by the scene depth characteristics, and the scene depth difference in a local area is relative to the scene of the entire image. The depth difference is indeed reduced, but the difference cannot always be ignored, and there are obvious scene depth differences in some local areas

The defogging algorithm of the dark channel principle assumes that the scene transmittance inside each local area is a constant, which leads to a block effect in the image after defogging

In order to solve the block effect problem in the prior art, it is necessary to further refine the local scene transmittance by using soft matting or guided filtering, but both soft matting and guided filtering need to traverse all the pixels of the image Performing pixel-by-pixel calculations consumes relatively large computing resources and is prone to long time delays. This is not suitable for low-cost hardware configurations in access control systems and application scenarios that require fast response.

In addition, although the dark channel principle is established with a high probability, it is not 100% applicable. For example, if there is a large white area in the image, it may cause the defogging algorithm of the dark channel principle to fail.

In addition, if there are individual abnormally bright points (such as reflective points under strong light) in the image, it may also cause a deviation in the estimation of the value of A in the above formula, which has to be considered in practical applications, especially in Among the devices that require high reliability, such as the access control system

Images