FPGA-Based Underwater Image Enhancement: Algorithmic Investigation and Hardware Deployment

Underwater image processing technology has become increasingly crucial in aquatic environmental monitoring and underwater equipment inspection, where edge-side algorithmic implementation forms the cornerstone for practical deployment. The underwater image algorithm suitable for the above purposes faces two problems: firstly, uneven distribution of underwater media can easily cause insufficient brightness and distorted underwater images, and secondly, optimization of algorithm deployment on the end side. Therefore, this article proposes an FPGA-based algorithm for underwater image enhancement - SCRE(Self-adaptive Correction and Retinex Enhancement). The algorithm integrates adaptive Gamma correction and Multi-Scale Retinex with Color Restoration (MSRCR). It establishes a coordinated “luminance optimization–color correction” enhancement framework in the HSV color space. Adaptive Gamma correction is applied to the Value (V) channel to enhance luminance effectively. Concurrently, during the color correction stage, a modified MSRCR method incorporating a compensation offset mechanism targeting underwater spectral attenuation is employed to perform color restoration and enhancement on the entire image.Furthermore, hardware optimization strategies achieved hardware deployment and algorithm acceleration based on the FPGA platform, pipeline architecture, and data parallelization techniques. Experimental verification shows that the SCRE algorithm improves UIQM and UCIQE metrics by 70.1%and 92.5%, respectively, compared to the original images, and deploying on FPGA hardware platforms, the processing speed of 1280×720 and 640×480 pixel images reaches 40fps/s and 70fps/s, respectively. Its processing speed is 9 to 10 times faster than a PC (Intel Core i7-14650H). Meeting real-time processing needs.