Images obtained in hazy environments present the phenomenon of low contrast, low saturation, and hue offset. Therefore, the main requirements for a dehazing system are accurate color restoration effects and high visual range. In this paper, an improved atmospheric scattering model based on elliptic curve is proposed to remove haze from a single image. Traditional atmospheric scattering model is improved in the following two aspects. Because of uneven illumination in hazy environment, scene incident light is improved by localized operation to replace the global atmospheric light. Considering multiple scattering effects of reflected light, the scene albedo is refined into two parts, the direct transmission item and the indirect transmission item. An elliptic curve model is established to adaptively estimate weights of the two terms. Thus, the reflected light under different conditions of depth and fog density is finely described. Experimental results show that our method achieves good visual performance. Compared with current methods, the indicator of image visibility and visual contrast increased by 40% and 86.6% on average.

在朦胧环境中获得的图像呈现出低对比度，低饱和度和色相偏移的现象。因此，除雾系统的主要要求是准确的色彩恢复效果和高视野。提出了一种改进的基于椭圆曲线的大气散射模型，以消除单幅图像中的雾度。传统的大气散射模型在以下两个方面进行了改进。由于朦胧环境中的照明不均匀，因此通过局部操作来代替全局大气光可以改善场景入射光。考虑到反射光的多重散射效应，场景反照率可分为直接透射项和间接透射项两部分。建立椭圆曲线模型以自适应地估计两个项的权重。因此，很好地描述了在不同深度和雾度条件下的反射光。实验结果表明，该方法具有良好的视觉效果。与现有方法相比，图像可见度和视觉对比度指标平均分别提高了40％和86.6％。