A Fast and Efficient Atmospheric Light Estimator for Underwater Image Dehazing

Atmospheric light and transmission estimations are the most important steps in underwater image dehazing based on dark channel prior. In this paper, we develop a fast and efficient method to estimate the atmospheric light according to the red channel prior for underwater image dehazing. The subsampling technique is first applied to reduce the computational complexity of atmospheric light estimation with almost no visual quality degradation. Accordingly, a low-cost VLSI architecture with heavy resource sharing for the atmospheric light estimation is proposed to meet the requirement of real-time underwater image dehazing. Compared to previous design, the proposed atmospheric light estimator requires much less hardware cost and can achieve 2.3 times speedup while maintaining good visual quality.