To address the issues of blurred boundaries, chamfered transitions, irregular geometries, and excessive dependence on empirical judgment in determining blast body model boundaries for 3D digital open-pit blasting design, this study proposes a tiered boundary identification and reconstruction methodology grounded in geometric feature analysis of the blast formations. Firstly, a multi-feature weighted fusion strategy was employed for top surface identification, systematically integrating four key geometric characteristics:normal vector orientation, slope gradient, circularity index, and horizontal alignment. This synergistic multi-criteria validation significantly improved detection precision. Secondly, a two-stage segmentation algorithm incorporating normal-vector alignment principles and spatial connectivity analysis was developed to classify side surfaces as contour faces or free faces. Finally, through the combined application of least-squares fitting, the Alpha Shape algorithm, and arc-length parameterization techniques, critical profile boundaries, including the bench crest line and toe line of the blast body geometry, were accurately reconstructed and mathematically optimized. Based on the above research findings, a boundary recognition system was successfully developed to accurately identify and reconstruct a 3D blast model. Practical validation using operational models from the Changlai mining area demonstrated the system′s stable performance and reliable recognition capabilities. Comparative analysis with manual identification demonstrated a 1.70% area deviation in top-surface recognition, 98% spatial overlap accuracy, a 1.94% bench-face line-length discrepancy, and a 0.56% relative root-mean-square error, confirming the method′s reliability and practical viability. This study contributes novel technical solutions for advancing digital transformation in open-pit blasting design.