基于软抠图指导性滤波的雾天图像复原技术

Fog Image Recovery Technology Based on Soft Matting Guiding Filter

在雾天环境下进行图像的采集具有大量的雾噪,雾天天气下的图像成像和图像远程识别效果不好,需要进行有效的雾天图像修复,对雾天图像的去雾复原技术是雾天图像处理的难题。传统方法采用盒子滤波法进行去雾复原,测量雾天图像三维重建相关参数具有不确定性,修复效果不好。提出一种基于软抠图指导性滤波的雾天图像复原技术。构建基于大气散射环境下的雾天图像采集模型,设计软抠图指导性滤波器实现对雾噪的合理滤波,对原始雾天图像进行雾化区域和非雾化区域的二值处理,提取多尺度加权平均Retinex特征,作为加权向量,指导软抠图滤波,实现雾天图像复原。仿真结果表明,采用该算法,进行雾天图像复原,能在除掉雾的同时,尽可能的保留了图像的本来颜色,提高信噪比,具有较好的雾天图像修复能力。

The image acquisition in the foggy environment with a large number of fog noise, fog weather image imaging and image remote recognition effect is not good, the need for effective image restoration of fog, fog image restoration technique is difficult to fog foggy image processing. The traditional method is using the box filtering method for restoration to fog, with uncertainty measurement of fog image 3D reconstruction of the relevant parameters, the repair effect is not good. Put for-ward a kind of guiding filtering of the fog image restoration technology based on soft matting. The construction of atmospher-ic scattering environment fog image acquisition model based on the design of soft matting, guiding filter to realize reason-able filtering noise of fog, the original fog image atomization region and non atomization region of two value processing, ex-tracting multi-scale weighted average Retinex characteristics, as the weighted vector, guide the soft matting filter, the real-ization of fog image restoration. The simulation results show that, using the algorithms of image restoration, fog, can get rid of fog in at the same time, as far as possible to retain the original color image, improve signal-to-noise ratio, has good fog image repairing ability.