基于大气散射模型的单幅雾天图像复原方法

Restoration for single fog image based on atmospheric scattering model

现有大气散射模型方法对于利用长期固定成像器的地方可以较好实现场景建模并复原雾天图像,但在缺乏其他天气状况下,景物图像作为参考的单幅雾天图像不能很好地复原。针对这一问题,该文提出一种新的基于大气散射模型的单幅雾天图像复原方法。该复原方法先通过边缘检测算法查找景物轮廓来划分不同区域,然后标定不同连通区域来划分景物层次,再计算图像灰度值变化曲线的斜率并分割天空区域,最后对新的环境光强度设定,对不同区域的景物采用不同的参数进行复原。采用一张手动拍射图像和一张网络图像作为实例分别进行实验,结果表明:该方法能够很好地复原缺乏参考图像的单幅雾天图像。

The current atmospheric scattering model method can be used to effectively realize scene modelling and restore fog image where the long -term fixed imager is used, but it is failed in restoring single fog image without reference of scene images under other weather conditions. Aiming at this problem, a new method of single fog image restoration based on atmospheric scattering model is presented in the article. The restoration method firstly can find the contours of the scenery to divide different regions in edge detection algorithm way, then divide the scene hierarchy by calibrating different regions together and calculate the slope of the image gray value change curve and divide the sky area, finally reset the n e w environment light intensity and restore sceneries in different regions by applying different parameters. A manually shot image and a network image are used as an example to test the method. The test results show that the method can effectively reconstruct the single fog image without reference images.