基于图像退化模型的天气现象识别

The Weather Phenomena Identification Based on Image Degradation Model

在不同的天气条件下,由于气溶胶对光波的散射作用,通过成像系统获得的图像都会有一定程度的退化。利用图像的退化模型可以获取与天气现象相关的参量,从而达到识别天气的目的。对不同天气条件下得到的同一场景图进行研究,运用多垂线检测法,检测出不同天气条件下直线模糊边缘上的灰度值,采用Sigmoid函数拟合出刃边函数,进而求出线扩散函数。通过分析线扩散函数的变化规律,总结出其与天气现象之间的紧密联系。测试结果表明,应用该方法对于晴天、薄雾、雨、霾、沙尘暴等天气现象具有一定的判别效果。

The images obtained from imaging system will have a certain degree of degradation due to the role of aerosols scattering on light under different weather conditions.The weather-related parameters can be acquired by using the degradation model of image and to achieve?the purpose of identifying the weather.In this paper,the same scene images obtained in different weather condition were studied,the method of multi-vertical detection was used to detect gray value on straight blur edge,and obtained the edge spread function fitted by Sigmoid function,then got the line spread function.Ultimately The relationship between weather phenomena and the line spread function is concluded by analyzing the variation of the line spread function.Test results show that the method presented in this paper have a certain discrimination effect for the sunny,mist,haze,rain, dust storms and other weather phenomena.