基于改进暗通道先验与自适应校正的水下图像复原增强

Underwater Image Restoration-Enhancement Algorithm Based on an Improved Dark Channel Prior and Adaptive Correction

  • 摘要: 针对水下图像受到光的吸收、反射、散射而导致的对比度低、颜色失真、细节模糊等问题,提出一种基于改进暗通道先验与自适应校正的水下图像复原增强方法。基于Jaffe-McGlamery模型利用引导滤波估计全局背景光。通过改进暗通道先验建立红、绿、蓝三通道透射率模型。在估计背景光和介质透射率得到复原图像的基础上,采用自动色阶与自适应伽马校正方法修正图像颜色信息。结合改进Sobel算子,在各通道独立HSI颜色空间内,对亮度分量进行多方向细节提升,以得到色彩信息丰富,符合人眼视觉感知的复原和增强水下图像。面向两种公开水下图像数据集的实验结果表明,所提算法与其他典型水下图像复原增强算法相比,能够更有效地消除色偏,更利于提升图像的对比度,并改善了图像清晰度,同时在UIQM值和UCIQE值两项客观指标上表现更好。

     

    Abstract: To address problems such as low contrast, color distortion, and detail blurring of underwater images caused by light absorption, reflection, and scattering, this study proposes an underwater image restoration and enhancement method based on an improved dark channel prior and adaptive correction. First, underwater background light estimation was established using a guided filter based on the Jaffe–McGlamery model. Second, the transmittance of the red-green-blue three-channel aqueous medium was calculated based on an improved dark channel prior. The automatic color method and adaptive gamma correction were then used for image restoration enhancement. Finally, by combining the improved Sobel operator, the image edge detail in the brightness component was enhanced with respect to the HSI color space to obtain a restoration-enhanced underwater image with rich color information that is consistent with human visual perception. The experimental results from two public underwater image datasets showed that our method effectively eliminated color deviations and improved the contrast, visibility, and clarity of the underwater image, with higher UIQM and UCIQE values than those of comparative methods.

     

/

返回文章
返回