Abstract: In multi-view 3D reconstruction tasks for engineering sites, optimizing the layout of the camera array is crucial for enhancing reconstruction quality and reducing computational costs. Addressing challenges such as high parameter coupling and conflicts among multiple optimization objectives in existing layout optimization methods, this paper proposes a camera array optimization method based on the Multi-Objective Coati Optimization Algorithm (MOCOA). By integrating neural network fitting functions and improving the global search and local optimization mechanisms, the method obtains Pareto-optimal solutions to achieve an optimal balance among completeness, accuracy, and computational cost. In a sandbox simulation scenario, the point cloud completeness increased by 74%, reprojection error decreased by 62%, and data utilization efficiency improved by 111%, demonstrating the effectiveness and potential value of the proposed method in engineering applications.