古建筑全景AI图像双角度偏振去雾仿真

Simulation of Dual Angle Polarization Defogging in AI Images of Ancient Architecture Panorama

由于古建筑全景AI图像受光照条件实时变化影响,在复杂纹理和高对比度区域,存在着多种密度和颜色的雾气,若不能及时提取图像的偏振特征,会直接降低图像的去雾效果。为此,提出古建筑全景AI图像双角度偏振去雾仿真。对获取的古建筑原始图像实施去噪处理,提升图像质量,建立图像大气散射模型;根据模型获取建筑图像双角度偏振图像,提取图像偏振特征,划分图像天空区域;采用暗通道先验原理获取图像无穷远大气光强,结合上述建立的大气散射模型,实现古建筑全景AI图像的去雾增强。实验结果表明,上述方法开展图像去雾时,去雾效果好、性能高。

Currently,the panoramic AI image of ancient architecture is affected by real-time changes in lighting conditions.In complex textures and high-contrast areas,there are various fogs with different densities and colors.If the polarization features of images cannot be extracted in time,the dehazing effect of the image will be directly re-duced.To address this issue,a dual-angle polarization defogging simulation for panoramic Al images of ancient archi-tecture was proposed.At first,the original image of ancient architecture was denoised to improve image quality.And then,an atmospheric scattering model was built.Based on the model,the dual-angle polarization image of ancient ar-chitecture was obtained.Meanwhile,the polarization features of the image were extracted.The sky region of the image was segmented.Moreover,the dark channel prior principle was used to calculate the infinite atmospheric light intensity of the image.Combined with the atmospheric scattering model,the defogging enhancement of panoramic AI images of ancient architecture was realized.Experimental results show that the proposed method has good dehazing effect and high performance.