Underwater computational ghost imaging LiDAR for multi-target detection with a super-low sampling ratio

This paper presents an underwater computational ghost imaging (UCGI) light detection and ranging (LiDAR) system that incorporates a novel, to our knowledge, wavelet transform-based Hadamard (WTH) ordering. Unlike traditional sequency-based orderings that often cause image elongation at low sampling ratios, WTH prioritizes modulation patterns based on their correlation with natural scene statistics, achieving balanced information capture in both spatial dimensions. This enables precise and robust multi-target detection at exceedingly low sampling ratios, even in highly turbid water. In Jerlov 9C water, under a sampling ratio of 10%, the system can achieve a superior performance of imaging, with a 97.58% reduction in mean squared error (MSE), a 55.24% increase in peak signal-to-noise ratio (PSNR), and a 528.92% enhancement in the structural similarity index (SSIM) compared to Sylvester Hadamard ordering. Furthermore, the system attains ranging precision and accuracy surpassing 3.90 and 6.40 mm, respectively. Thus, WTH provides a promising solution for high-speed imaging and ranging in dynamic environments, especially underwater.