基于大气散射模型的偏振图像去雾方法

雾霾天气越来越常见,导致所采集图像的应用价值降低,因此如何获得清晰度高的图像成为计算机视觉等领域重点研究的内容。在大气散射模型的基础上提出一种利用图像偏振信息的去雾方法。该方法首先利用偏振成像系统获取平行和垂直方向的两幅偏振图像,结合暗原色先验理论自动估计大气光强信息,估计传输率图并采用改进的导向滤波的方法优化传输率图,最终实现图像去雾。实验结果表明,该算法提高了图像的清晰度及对比度,有效改善了雾天条件下景物的视觉效果。

Polarization Image Defogging Algorithm Based on Atmosphere Scattering Model

Haze weather is more and more common. It has resulted in reducing the application value of the collected image. So how to obtain high-definition images becomes the research focus in the field of computer vision and other areas. Based on atmospheric scattering model, an image defogging method using image polarization information is pro-posed. The method firstly acquires two polarization images in the parallel and vertical direction using polarization imag-ing system. Then automatically estimates the atmospheric light intensity information and transmission rate combined with the dark channel prior. And guided filter method is used to optimize transmission rate,ultimately realizes image defog-ging. Experimental results have shown that the algorithm improves the image’s definition and contrast. It effectively im-proves scenery visual effect in the haze condition.