基于雾天图像退化模型的自适应参数优化的去雾算法

针对单幅雾霾图像中包含有大面积浓雾、高亮以及白色物体等,而导致无法清晰识别的问题,基于雾天退化模型,提出了一种改进暗通道和运用灰度开运算求解环境光值相结合的去雾算法。首先根据暗通道先验理论运用图像阈值分割出暗原色区域和明原色区域,并将暗原色区域与明原色区域相结合以求得更加精准的原始透射率;然后采用引导滤波算法细化原始透射率;并通过灰度开运算对环境光值进行区间估计,提高了环境光值的精准性和鲁棒性。使得该算法适用于暗通道去雾效果不好的浓雾高亮区域,去雾后的图像更加真实自然,边缘细节信息更加丰富,有效去除了Halo效应;同时也有效地解决了单幅图片去雾后图片偏暗,图片视觉效果不好等问题。与经典去雾算法作比较,验证在图像的对比度、失真度、细节信息和边缘保持等方面都优于其他算法。

De-fogging Algorithm Based on Adaptive Parameter Optimization for Foggy Image Degradation Model

For a single smog image, there are some problems such as large area of thick fog, high brightness, and white objects. Based on the foggy day degradation model, a defogging algorithm is proposed to improve the dark channel and solve the ambient light value by using the gray-scale open operation in this paper. Firstly, according to the dark channel prior theory, the image threshold is used to segment the dark primary color region and the bright primary color region.And the dark primary color region and the bright primary color region are combined to obtain a more accurate original transmittance. Then, the guided filter algorithm is used to refine the original transmittance, And the interval of the ambient light value is estimated by the gray-scale operation to improve the accuracy and robustness of ambient light values. Make the algorithm suitable for dense fog highlighting area with bad dark channel defogging effect.The image after defogging is more real and natural, and the edge details are more abundant, effectively removing the Halo effect.At the same time, it also effectively solves the problem that the picture is dark after defogging a single picture, and the visual effect of the picture is not good. Comparing this algorithm with the classic defogging algorithm,it also directly proves that the processed image is superior to other algorithms in terms of contrast, distortion, detail and edge retention.