基于小波分解和饱和度估计的水下图像复原

Underwater image recovery based on wavelet decomposition and saturation estimation

在水下成像时,由于受光线吸收、前向散射和后向散射等因素的影响,从而导致水下图像出现严重退化的现象.针对这些问题,提出了一种利用小波分解和饱和度估计透射率的水下图像复原方法.首先基于水下图像的频谱特征,借助小波分解,在最低频子带利用饱和度估计透射率来抑制散射光,在高频子带抑制噪声并进行细节增强.然后利用自动白平衡和饱和度增强的直方图匹配来对水下图像进行增强,从而改善水下图像的颜色偏差和对比度偏差.评估指标NIQE和UCIQE的均值分别为3.2177和0.4895,优于其他算法.实验结果表明:这种方法可以得到颜色更自然,饱和度更均衡的复原图像.

In underwater imaging,the underwater image is seriously degraded due to the influence of light absorption,forward scattering and backward scattering.This paper proposes an underwater image restoration method based on wavelet decomposition and saturation estimation of transmittance.Firstly,based on the spectral characteristics of underwater images,the scattering light is suppressed by estimating transmittance with saturation in the lowest frequency subband.The noise is suppressed and the details are enhanced in the high frequency subband with the help of wavelet decomposition.Secondly,automatic white balance and saturation enhancement histogram matching are used to enhance the underwater image,so as to improve the color deviation and contrast deviation of the underwater image.The average values of NIQE and UCIQE are 3.2177 and 0.4895 respectively,which are better than other algorithms.Experimental results show that this method can get more natural color and more balanced saturation of the restored image.