Acta Photonica Sinica, Volume. 49, Issue 3, 0310003(2020)
Underwater Optical Image Enhancement Based on Dominant Feature Image Fusion
Aiming at the problems of color distortion, uneven illumination and low contrast of underwater optical image, an underwater optical image enhancement algorithm based on the fusion of dominant feature image was proposed. Firstly, an improved dark channel prior algorithm was proposed to remove the uneven turbidity and balance the color in the degraded image. Secondly, the adaptive gamma correction algorithm based on weighted distribution and the contrast limited adaptive histogram equalization-homomorphic filtering algorithm were used to enhance the contrast of color correction image and make its brightness distribution uniform. Finally, the associated weight maps of the three fused images namely the color-corrected image, the brightness-balanced image and the contrast-enhanced image were defined, and the fused images were obtained by the multi-scale fusion algorithm. Compared with single preprocessing algorithm which can only solve the corresponding degradation phenomenon, the algorithm processes single degraded image with multiple algorithms and obtains three dominant feature images, the combination of different weights can combine the dominant features to the greatest extent, and the comprehensive effect is far beyond the optimization effect of each single algorithm, and is no longer limited to solving single problems such as color distortion. The algorithm in this paper is compared with existing algorithms in subjective evaluation and objective evaluation. The results show that the algorithm can effectively balance the chroma, saturation and sharpness of underwater images, and the visual effect is close to the images in natural scenes.
Get Citation
Copy Citation Text
Sen LIN, Kai-chen CHI, Wen-tao LI, Yan-dong TANG. Underwater Optical Image Enhancement Based on Dominant Feature Image Fusion[J]. Acta Photonica Sinica, 2020, 49(3): 0310003
Category: Image Processing
Received: Sep. 24, 2019
Accepted: Dec. 27, 2019
Published Online: Apr. 24, 2020
The Author Email: