eduzhai > Applied Sciences > Engineering >

Low-Light Maritime Image Enhancement with Regularized Illumination Optimization and Deep Noise Suppression

  • king
  • (0) Download
  • 20210506
  • Save

... pages left unread,continue reading

Document pages: 17 pages

Abstract: Maritime images captured under low-light imaging condition easily suffer fromlow visibility and unexpected noise, leading to negative effects on maritimetraffic supervision and management. To promote imaging performance, it isnecessary to restore the important visual information from degraded low-lightimages. In this paper, we propose to enhance the low-light images throughregularized illumination optimization and deep noise suppression. Inparticular, a hybrid regularized variational model, which combines L0-normgradient sparsity prior with structure-aware regularization, is presented torefine the coarse illumination map originally estimated using Max-RGB. Theadaptive gamma correction method is then introduced to adjust the refinedillumination map. Based on the assumption of Retinex theory, a guidedfilter-based detail boosting method is introduced to optimize the reflectionmap. The adjusted illumination and optimized reflection maps are finallycombined to generate the enhanced maritime images. To suppress the effect ofunwanted noise on imaging performance, a deep learning-based blind denoisingframework is further introduced to promote the visual quality of enhancedimage. In particular, this framework is composed of two sub-networks, i.e.,E-Net and D-Net adopted for noise level estimation and non-blind noisereduction, respectively. The main benefit of our image enhancement method isthat it takes full advantage of the regularized illumination optimization anddeep blind denoising. Comprehensive experiments have been conducted on bothsynthetic and realistic maritime images to compare our proposed method withseveral state-of-the-art imaging methods. Experimental results have illustratedits superior performance in terms of both quantitative and qualitativeevaluations.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×