欢迎访问《兵工学报》官方网站,今天是

兵工学报

• •    下一篇

基于神经辐射场和层次分析的坦克三维重建和毁伤评估研究

张子豪1,娄文忠1*,郭志明1,2,赵飞1,丁男希1,李成龙1,马文龙1   

  1. 1.北京理工大学 机电学院; 2.中国兵器科学研究院
  • 收稿日期:2024-12-26 修回日期:2025-03-11

Research on Tank Damage Assessment Based on Neural Radiance Fields and Hierarchical Analysis

ZHANG Zihao1,LOU Wenzhong1 *, GUO Zhiming1,2, ZHAO Fei1,DING Nanxi1,LI Chenglong1,MA Wenlong1   

  1. 1. School of Mechanical Engineering, Beijing Institute of Technology; 2.China Ordance Science Institute
  • Received:2024-12-26 Revised:2025-03-11

摘要: 随着现代战场环境日趋复杂,实时、准确地评估目标毁伤效果已成为提升作战效能和指挥决策的重要前提。针对战场环境下目标的毁伤效果评估问题,提出一种基于神经辐射场和层次分析法的坦克毁伤评估模型,通过结合分割一切模型和神经辐射场技术,实现对坦克各部件的精细分割和三维重建。构建一个包含机动性能(车体、履带)和攻击性能(炮管、炮塔)的坦克毁伤树模型,全面反映坦克在战场中的功能损失,并利用匈牙利算法进行部件与毁伤区域的匹配,采用三角模糊层次分析法对匹配结果进行量化分析,精确评估毁伤等级。实验结果表明,新方法在目标分割和三维重建方面优于传统方法,尤其在复杂战场环境下表现出显著优势,突破了传统基于二维信息毁伤评估的局限性,显著提高毁伤评估的准确性和可靠性。

关键词: 坦克, 毁伤评估, 三维掩码, 神经辐射场, 匈牙利算法, 层次分析法

Abstract: As modern battlefield environments become increasingly complex, real-time and accurate assessment of target damage has become a vital prerequisite for enhancing combat effectiveness and informed command decisions. To address the challenge of damage assessment for battlefield targets, a tank damage assessment model based on Neural Radiance Field (NeRF) and the Analytic Hierarchy Process (AHP) is proposed. By integrating the Segment Anything Model (SAM) with NeRF, precise segmentation and 3D reconstruction of tank components are achieved. Additionally, a tank damage tree model is constructed, incorporating mobility performance (hull, tracks) and attack performance (gun barrel, turret), providing a comprehensive representation of functional losses of tanks on the battlefield. The Hungarian algorithm is applied to match components with damage areas, and the triangular fuzzy AHP is employed to quantify the matching results, enabling accurate damage level assessment. Experimental results demonstrate that this method outperforms traditional approaches in target segmentation and 3D reconstruction,, especially in complex battlefield environments. It breaks through the limitations of traditional 2D-based damage assessment, significantly improving the accuracy and reliability of damage evaluation.

Key words: tank, damage assessment, 3D mask, neural radiance fields, hungarian algorithm, analytic hierarchy process

中图分类号: