结合自适应亮度变换不等式估计透射率的图像去雾方法

An Image Dehazing Method Combining Adaptive Brightness Transformation Inequality to Estimate Transmittance

为了解决单幅图像去雾算法中透射率估计不准确和去雾不彻底问题,提出了一种结合自适应亮度变换不等式估计透射率的图像去雾方法。首先,利用亮度和反饱和度进行亮度变换不等式补偿构造场景深度,并使用对数变换进行灰度扩展;其次,采用正则化对场景深度优化,解决相邻像素的差异过大问题,逼近精确场景深度;然后,根据局部恒定假设,利用局部平均深度和联合双边滤波估计动态大气散射系数,结合场景深度和动态大气散射系数获得透射率;最后,使用中值滤波和区间估值优化局部大气光,并通过大气散射模型恢复无雾图像。实验结果表明:所提出的图像去雾方法可在不同场景自适应地调节模型参数,去雾彻底,获得了较好的去雾效果;同时,客观指标中平均视觉对比度达到了60.572,具有较好的图像保真度。

An image dehazing method combining adaptive brightness transformation inequality to estimate transmittance is proposed to solve the problems of inaccurate transmittance estimation and incomplete dehazing in single image dehazing algorithms.Firstly,the brightness transformation inequality is applied to construct the scene depth by using brightness and inverse saturation,and a logarithmic transformation is used to expand gray scale.Secondly,regularization is used to optimize scene depth to solve the excessive difference between adjacent pixels and to approximate the exact depth of the scene.Then,according to the local constant assumption,the dynamic atmospheric scattering coefficient is estimated using local average depth and joint bilateral filtering,and the transmittance is obtained by combining the scene depth and dynamic atmospheric scattering coefficient.Finally,the median filtering and interval estimation are used to optimize the local atmospheric light,and clear image is recovered by the atmospheric scattering model.Experimental results show that the proposed method can adaptively adjust the model parameters in different scenes,thoroughly remove haze,and get better dehazing effect.At the same time,the average visual contrast in objective indexes reaches 60.572,with good image fidelity.