Abstract: At present, simulation visualization methods based on the augmented reality (AR) of high arch dam construction have drawbacks in limited visualization angles and poor system integration. Mobile AR has the advantage of diversified interaction modes but suffers relatively large errors in camera registration. This paper presents a new mobile AR method for visualizing high arch dam construction simulations based on optimization via simultaneous localization and mapping (SLAM). This new visualization system calculates accurately the initial camera registration parameters based on SLAM and the 3D and 2D imaging relationship of the target, and realizes real-time camera registration through ARCore motion tracking. Application to high dam construction shows that it improves the overall accuracy in initial camera registration by 97.6% against the traditional sensor method, and its average error in mobile camera registration is only 0.275 m, thus enabling its success in multi-directional spatial decision analysis.

Key words: high arch dam, construction simulation visualization, mobile AR, SLAM, ARCore