Research On Defogging Fusion Method Of Color And Polarization Images

Haze weather will have a serious impact on monitoring,security and other fields.Traditional defogging methods can improve the clarity of fog image.In the case of local high exposure and limited sky area,the defogging effect is not ideal.At the same time,the traditional defogging model takes the atmospheric light value constant as the prior condition,which leads to the loss of near detail information.Based on the above analysis and the atmospheric degradation model,this paper studies the robustness of the defogging method in complex scenes and the details enhancement of the defog image.1)To solve the problem of inaccurate solution of fog parameters in complex scenes,a method of estimating atmospheric light intensity at infinity based on atmospheric model fog removal is designed in this paper.First of all,based on the characteristic that polarization difference image seldom contains the target reflected light component,polarization difference image is introduced to be the four binary tree index target image and restrain the influence of high exposure point.In order to improve the retrieval capability of the limited sky area and reduce the index time overhead,the quadtree structure is optimized with the pixel intensity scoring mechanism.Thus,the region of infinite atmospheric light intensity is determined,and the value of infinite atmospheric light intensity is confirmed in the corresponding region of the total polarized light intensity image.Secondly,the optical component is calibrated by polarization azimuth,which makes the atmospheric polarization degree and atmospheric light intensity more accurate Finally,the defogging image is obtained by the inverse solution of the atmospheric model.Finally,the defogging image is obtained by the inverse solution of the atmospheric model.The experimental results show that the method can resist the interference of high exposure point and limited sky application scene.Compared with traditional defogging,the objective evaluation criteria for defrosted images are improved by 21.3%and the contrast by 19.7%.2)Aiming at the problem that the detail of defogging image is not enough.An improved stimulus input based on SCM network model is designed as a high frequency fusion rule.The defogging image is fused with the polarization image,which highlights different material boundaries and improves the sharpness of the target in shadows.Firstly,in order to improve the color representation of the defrosted image,the color image saturation is enhanced adaptively based on the HSV color domain,and the improved guided filter and Gaussian filter are selected as the decomposition tools in the fusion process.Secondly,the average energy fusion rule is used for low frequency information.For high frequency information,detail fusion is performed using pulsed cortical SCM,and improved Laplace and edge energy are selected as external stimuli for SCM models corresponding to polarization and fog removal images,respectively.Finally,the fused result image is obtained by inverse scale transformation.The experimental results show that the algorithm effectively enhances image details and edges,and improves spatial frequency by 11.41%and Q_AB/Fby 12.5%compared with other fusion algorithms.Therefore,the defogging fusion method proposed in this paper based on color and polarized images has made effective technical extensions in fog image processing.