一种基于多尺度Retinex算法的图像去雾方法

摘要

Images taken in foggy weather are blurred in view and with poor contrast because of atmosphere scattering, these seriously impact the application of traffic system and outdoor visions system. By improving the degradation model of atmospheric physics and based on multi-scale Retinex (MSR) algorithm, we propose a new defogging method. This method sets up the model for images according to the imaging mechanism in foggy weather, and then processes the images being modelled based on MSR algorithm. Experiments demonstrate that the method can effectively eliminate the foggy impact on the image and realises the recovery of colour degraded images.%在有雾天气条件下拍摄的图像,由于大气的散射作用,导致图像的内容模糊不清,对比度下降,给交通系统及户外视觉系统的应用带来严重的影响.通过改进大气物理退化模型,在多尺度Retinex算法MSR(Multi-Scale Retinex)的基础上,提出一种新的去雾方法.该方法根据雾天成像机理对图像建立模型,再根据MSR算法对建模后的图像进行处理.实验表明,该方法能有效去除雾化效果,实现彩色退化图像的复原.