基于线性加权的自适应图像去雾算法

THE SELF-ADAPTIVE IMAGE DEHAZING ALGORITHM BASED ON LINEAR WEIGHTING

在雾霾天气下,大气散射作用导致采集图像信息丢失。针对这一问题,以暗通道先验原理为基础,提出一种基于线性加权的自适应图像去雾算法。首先,在计算暗通道函数时,采用一种改进方法生成精确的暗原色图,并使用图像锐化技术确保场景边界特性;其次,针对复原图像对比度过深,定义一种自适应的线性加权方式计算准确的大气光强值,确保得到代表实际场景的透射率图;最后基于大气散射的物理模型,得到清晰的无雾复原图片。实验结果表明,该方法能有效地实现图像去雾,且具有效果好和速度快的优点。

In haze weather, the effect of atmospheric scattering lead to the missing of the self-adaptive image dehazing algorithm based on linear weighting is proposed on the channel prior. Firstly, during the calculation of dark channel function, the proposed method to generate the accurate dark primary picture, and uses the image sharpening the image information. Thus, basis of the principle of dark approach adopts an improved technique keep the boundary characteristics of the object. Then, aiming at the over-high contrast of recovered image, a self-adaptive linear weighting method is defined to calculate the accurate values of atmospheric light intensity, ensuring that the transmittance picture of actual scene is obtained- The.experimental results show that the proposed approach can reliably achieve image dehazing with good quality and efficiency.