Journal
APPLIED INTELLIGENCE
Volume 52, Issue 14, Pages 16435-16457Publisher
SPRINGER
DOI: 10.1007/s10489-022-03275-z
Keywords
Underwater images; Multi-feature prior; Artificial exposure; Color correction
Categories
Funding
- National Natural Science Foundation of China [61702074]
- Liaoning Provincial Natural Science Foundation of China [20170520196]
- Fundamental Research Funds for the Central Universities [3132019205, 3132019354]
Ask authors/readers for more resources
This paper proposes a visual quality enhancement method for underwater images based on multi-feature prior fusion (MFPF), which extracts and fuses multiple feature priors to enhance the visual quality of underwater images.
The information in a single underwater image is insufficient due to the complexity of the underwater environment, which makes it challenging to meet the expectations of marine research. In this paper, we proposed a visual quality enhancement method for underwater images based on multi-feature prior fusion (MFPF), achieved by extracting and fusing multiple feature priors of underwater images. Complementary multi-features enhance the visual quality of underwater images. We designed a color correction method based on self-adaptive standard deviation, which realizes the color offset correction based on the dominant color of the underwater image. A gamma correction power function and spatial linear adjustment were also applied to achieve a set of artificial exposure map sequences obtained from a single degraded image and enhance the dark area's brightness and structural details. This design makes full use of the advantages of white balance, guided filtering, and multi-exposure sequence technology. And it uses a multi-scale fusion of various prior features to enhance underwater images. The experimental results show that by applying the multi-feature prior fusion scheme, this design comprehensively solves various degenerated problems, removes over-enhancement, and improves dark details.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available