Imaging Through Smoke Under Strong Light Interference

In smoke scenes without strong light interference,image enhancement,image restoration,deep learning and other methods can effectively recover the hidden object information in the smoke.When there exists strong light interference in the imaging scene,both the smoke scattering and the strong light will cause the object information to be submerged by the strong light,and the object is difficult to be observed and reconstructed.To solve this issue,this thesis studies the influence of atmospherictransmittance ratio,light intensity and glare angle on the reconstruction of object images in the strong light smoke scene.On this basis,the smoke image is separated according to the smoothness difference between the smoke image and the strong light interference information.After preprocessing,the object image is reconstructed based on Retinex algorithm.The research in this paper has important practical significance for the detection and imaging of targets in strong light and smoke scenes such as fire rescue,battlefield reconnaissance,maritime detection,disaster warning and foggy days driving.The specific contents of this paper are depicted as follows:.(1)The influence of strong light interference in smoke imaging.On the basis of analyzing the degradation mechanism of foggy images,dark channel prior,homomorphic filtering and histogram equalization are selected as three common foggy removal algorithms to explore the change rules of images recovered from strong light smoke scenes.The effects of atmospheric transmittance ratio,light intensity and glare angle on smoke images are investigated based on the simulation.When the atmospheric transmittance ratio is lower than0.4,the three algorithms failed for smoke images reconstruction without introducing strong light.The reconstructed image quality and effective recovery distance are reduced at the same time after the addition of strong light,and the interference of strong light to image reconstruction is the most serious in the range of atmospheric transmittance ratio 0.4-0.55.While the angle between the injection of strong light and imaging optical axis is 30°and 90°,the reconstructed image quality in the direction of upper surface of vertical smoke bin varies greatly with the change of illumination intensity,and the order of its influence on the attenuation of image restoration quality from strong to weak is 30°,90°,and the direction of upper surface of vertical smoke bin.(2)Separation and reconstruction of strong light smoke images.Based on the traditional atmospheric scattering model combined with the transmission law of strong light and the information distribution of strong light smoke images,the atmospheric scattering model under strong interference light is established.On this basis,according to the difference in the smoothness of the information between the smoke image and the strong light,the distribution law on the gray gradient is obtained,and the separation and extraction of the strong light smoke image is transformed into the problem of finding the maximum joint probability distribution.After acquiring the separated image,the median filter and grayscale Gamma correction algorithm were used to make the image closer to the smoke image without glare interference,then the best correction parameter distribution interval 1.2-1.4was determined based on the reconstructed image quality results,and finally the Retinex algorithm was used to realize the reconstruction of glare smoke image.Compared with the traditional smoke transparent algorithm,the strong light smoke image reconstructed by this method is improved by up to 25%,AG by up to 80%,and SF by up to 120%,and the visual effect is significantly improved.