Abstract:

The rapid development of artificial intelligence has made image processing technology widely used in the new generation of intelligent transportation systems.The issue of insufficient estimation of the transmission map when the existing image dehazed algorithms are applied to Intelligent Transportation Systems leads to the color shift,artifacts,and low contrast in the sudden depth of the field area for the restored images and seriously affects the performance of the outdoor acquisition system.Therefore,this paper proposes an adaptive transmittance defogging algorithm based on a nonlinear transformation.Through the use of logarithmic transformation and adaptive parameters,the intensity value of the high gray area in a dark channel is compressed that can obtain the dark channel of the original fog-free image.And then the initial transmittance is estimated.According to the difference between pixel brightness and saturation,an adjustment factor is introduced to compensate the transmittance of the sky area.After that,by combining with guided filtering,the compensated transmittance is smoothed to obtain the adaptive optimized transmittance.Then,on the basis of the atmospheric scattering model,the dehazed results are obtained.Simulation results show that the algorithm has a clear and natural dehazed effect on the sky and in the sudden depth of field area,with rich texture details,no obvious artifact and color shift,and moderate brightness.We conduct extensive experiments to report quantitative results for comparison,such as average gradient,signal-to-noise ratio,structural similarity,and information entropy.The parameters are better than those of other linear algorithms,and each index is improved by about 6.4% on average,which can effectively alleviate the halo and distortion of the dehazed image in the sudden depth of the field area.

Key words: image processing, image dehazing, attenuation prior, logarithmic transformation