Abstract Deep water, distant sea, unmanned is the inevitable trend of the development of marine engineering, the underwater positioning system for the accuracy, real‐time, and environmental adaptability of the aspects of the increasingly high requirements. The mainstream underwater positioning methods face limitations such as multipath effects, cost, water depth, and water quality, making it difficult to meet diverse needs. This study presents a novel underwater photogrammetry solution based on an active cooperative target that combines optical hardware with intelligent algorithms to achieve millimeter‐level positioning in complex marine environments. Specifically, the system designs and optimizes the hardware configuration, including binocular vision camera, LED array target, and auxiliary optics, through multi‐parameter association to ensure the continuity and stability of positioning. At the algorithmic level, a multilevel image processing module is established through spatiotemporal distribution analysis, expected template matching, physical light intensity modeling, and geometric configuration constraints, which effectively overcomes the dynamic occlusion, scattering degradation and feature extraction errors of cooperative targets. In a standard test cell, the system achieves an angular accuracy of 0.24° and a ranging accuracy of 0.72 mm. A number of positioning systems have been developed to assist in the docking of submarine immersed tube tunnels, and the absolute positioning error is still better than 5 mm even under dynamic high turbidity conditions, which proves its effectiveness.
Ma et al. (2025) studied this question.