Computer Engineering and Applications ›› 2022, Vol. 58 ›› Issue (11): 224-233.DOI: 10.3778/j.issn.1002-8331.2009-0497

• Graphics and Image Processing • Previous Articles     Next Articles

Adaptive Retinex Algorithm Based on Detail Selection Used in Underwater Image Enhancement

XU Li, LU Guiming, QIU Zhenguang   

  1. School of Physics and Electronics, North China University of Water Resources and Electric Power, Zhengzhou 450046, China
  • Online:2022-06-01 Published:2022-06-01



  1. 华北水利水电大学 物理与电子学院,郑州 450046

Abstract: Aiming at the contradiction of color distortion and image detail enhancement when Retinex algorithm is applied to underwater image enhancement, the adaptive multi-scale Retinex underwater image enhancement algorithm combined with detail information is proposed in this paper. The selection requirements of the scale of convolution function in the enhancement of Retinex algorithm for underwater images containing different detail information are analyzed. Image gradient is used as the adjustment factor to adaptively adjust the weight of multi-scale Retinex operator, which is used to meet the requirements of underwater image with different detail information for contrast enhancement, and effectively alleviate the contradiction between color distortion and detail contrast enhancement in underwater image enhancement. Several groups of experiments show that the algorithm is superior to the traditional multi-scale(MSR) and multi-scale Retinex with color restoration(MSRCR) in removing the blue-green background of underwater image, avoiding color distortion, eliminating non-uniform illumination and image detail enhancement.

Key words: underwater image enhancement, Retinex, multi-scale Retinex(MSR), adaptive multi-scale Retinex(AMSR), image gradient

摘要: 针对Retinex算法应用于水下图像增强中,常出现颜色失真与图像细节增强相矛盾的现象,提出了结合细节信息的自适应多尺度Retinex水下图像增强算法。分析包含不同细节信息的水下图像对Retinex算法增强中卷积函数尺度大小的选择要求;采用图像梯度作为调节因子,自适应调整多尺度Retinex算子的权重,用于适应包含不同细节信息的水下图像对对比度增强的要求,有效地缓和了水下图像增强在颜色失真和细节对比度提升之间的矛盾。多组实验验证了该算法在去除水下图像的蓝绿背景、避免颜色失真、消除非均匀光照和图像细节增强等方面均优于传统多尺度和颜色保真的多尺度Retinex算法。

关键词: 水下图像增强, Retinex算法, 多尺度Retinex算法, 自适应多尺度Retinex算法, 图像梯度