Abstract: The volume and morphological parameters of the opencast mine blast pile are crucial indicators for evaluating the blasting effect. However, current methods for determining these parameters suffer from issues such as low accuracy, high operational costs, and slow speed. To promptly and accurately obtain the volume and other morphological parameters of the blast pile, an intelligent calculation method is proposed for the opencast mine blast pile volume based on 3D visual point clouds. Firstly, unmanned aerial vehicles (UAVs) are used to collect image datasets of the blast pile scene. Secondly, an opencast mine blast pile 3D reconstruction model is constructed based on Instant-NGP. Colmap is employed to obtain camera intrinsic parameters and image pose information, which are then used as inputs to the Instant-NGP model. The model is trained to output the raw point cloud data of the blast pile scene. Subsequently, a grid model-based method for calculating the opencast mine blast pile volume is proposed. The raw point cloud data is processed through point cloud filtering and scale recovery methods. The point clouds of the blast pile's base and surface are obtained using the Cloth Simulation Filter (CSF) algorithm and the Angle Criterion (AC) algorithm, respectively. A complete point cloud of the blast pile is acquired by downward-growing the point cloud, and geometric algorithms are used to obtain morphological parameters such as the throw distance of the blast pile. Finally, the alpha-shape surface reconstruction algorithm is applied to reconstruct the blast pile, yielding a mesh model. The volume of the blast pile is calculated by summing the tetrahedra in the mesh model. The analysis of experimental results indicates that the intelligent calculation method based on 3D visual point clouds accurately determines morphological parameters such as the throw distance, heap height, and width of the blast pile, with experimental errors all below 5%. Additionally, cross-sectional views of the blast pile are obtained using the slicing method. When comparing the calculated blast pile volume to the actual volume, the error is within 5%. Furthermore, the entire process of 3D reconstruction and point cloud processing proposed takes less than 25 minutes, meeting the demand for rapid acquisition of blast pile volume.