International Open Access Journal Platform

基于3D高斯泼溅技术的犯罪现场重建方法研究

Research on Crime Scene Reconstruction Methods Based on 3D Gaussian Splatting Technology

付文波

Progress in Social Sciences / 2026,8(5): 458-462 / 2026-06-01 9 5

View Download PDF

• Information:

  中央司法警官学院，保定

• Keywords:
  Scene reconstruction; 3D gaussian splatting; Methodology
  现场重建; 3D高斯泼溅; 方法
• Abstract: Technological advancements have fundamentally transformed the methods and approaches used in crime scene reconstruction. Technologies such as Virtual Reality (VR), 3D modeling, neural rendering, and Artificial Intelligence (AI) are being integrated into crime scene reconstruction to enhance precision and efficiency, addressing the limitations of traditional methods, including insufficient accuracy, low efficiency, and strong subjectivity. This paper focuses on research on crime scene reconstruction methods based on 3D Gaussian Splatting (3D GS) technology. It systematically reviews the current research status of crime scene reconstruction both domestically and internationally, providing a comparative analysis of the advantages and disadvantages of mainstream digital reconstruction methods, including traditional 3D reconstruction, 3D laser and structured light scanning, Neural Radiance Fields (NeRF), and LiDAR SLAM combined with 3D GS. The paper elucidates the core principles of 3D GS technology and identifies its technical advantages. It constructs a complete workflow for 3D GS-based crime scene reconstruction, encompassing five stages: data acquisition, data preprocessing, sparse point cloud generation and camera localization, model training and optimization, and scene reconstruction application and evidence presentation. Finally, the study analyzes three core challenges faced in practical application: adaptation to complex scenes, the geometric precision gap, and the insufficiency of judicial probative value. This research aims to improve the technical framework of digital crime scene reconstruction and provide theoretical support and reference for the intelligent transformation of criminal scene investigation. 技术的进步让犯罪现场重建方式方法发生巨大改变，基于虚拟现实技术、3D建模、神经渲染技术、人工智能技术正被应用于犯罪现场重建，以提升现场重建的精度和效率，解决传统犯罪现场重建方法精度不足、效率低、主观性强等问题。本文开展基于3D高斯泼溅技术的犯罪现场重建方法研究，系统梳理国内外犯罪现场重建研究现状，对比分析传统三维重建、三维激光和结构光扫描技术、NeRF及LiDARSLAM+3DGS等主流数字化重建方法的优劣；阐释3DGS技术的核心原理，明确其技术优势，构建基于3DGS的犯罪现场重建完整流程，涵盖数据采集、数据预处理、稀疏点云生成与相机定位、模型训练与优化、现场重建应用与证据呈现五个步骤；最后，分析该方法在实际应用中面临的复杂现场适配、几何精度鸿沟、司法采信与认证标准尚不完善三大核心挑战。旨在完善数字化犯罪现场重建技术体系，为犯罪现场勘查智能化转型提供理论支撑与参考。
• DOI: 10.35534/pss.0805080 (registering DOI)
• Cite: 付文波．基于3D高斯泼溅技术的犯罪现场重建方法研究［J］．社会科学进展，2026，8（5）：458-462.