Underwater enhancement computing of ocean HABs based on cyclic color compensation and multi-scale fusion

The clarity of algal images is crucial for solving problems such as harmful algal blooms (HABs) classification and red tide warning. However, the original underwater microscopic images of HABs often produce color distortion and blurred details due to the effects of seawater quality, illumination and image extraction techniques. To deal with the above problems, an enhancement computing method based on cyclic color compensation and three-image Multi-scale fusion is proposed in this paper. In this method, first, we propose a novel cyclic color compensation method to correct the color of the microscopic images of HABs. Then, with the input of three images derived from the color compensation versions of the original images of the algae, the correlation weight maps of the three inputs are redefined to facilitate the transfer of texture features and color contrast to the output images. Finally, to avoid image artifacts in the low-frequency components of the output image after the weight map transformation, a novel image Multi-scale fusion strategy is used in this paper. Qualitative and quantitative experimental results show that the underwater microscopic images of HABs enhanced with the method proposed in this paper have the best texture clarity and global contrast, and the method significantly improves the accuracy of transmission map estimation, edge detection and key point matching in processing algal images.