基于光照分量的多尺度水下图像增强

Multi-Scale Underwater Image Enhancement Based on Illumination Component

摘要: 针对传统方法中水下图像透射率估计不够精确的问题，本文提出了一种基于光照分量的多尺度优化估计方法。该方法基于Retinex理论与水下光学成像模型，认为图像的光照分量等价于透射率分量；并结合图像由细节部分与平滑部分构成原理，提出了比例融合方法，使透射率图像更加接近于实际；最后采用高斯金字塔，对透射率图像进行多尺度优化。实验结果表明，该方法能够较好的解决水下图像模糊、对比度低等问题。

Abstract: Aiming at the problem that the transmittance estimation in traditional underwater image en-hancement method is not accurate enough, this paper proposes a multi-scale optimization method based on illumination component. The method is based on Retinex theory and underwater optical imaging model, which thinks that the illumination component of the image is equivalent to the transmittance component; and combines the principle of detail and smoothness of the image to form a proportional fusion method, which makes the transmittance image closer to the actuality. Finally, the Gaussian pyramid is used to optimize the multi-scale of the transmittance image. The experimental results show that the method can solve the problems of blur underwater image and low contrast.