机电工程技术2025，Vol.54Issue(11)：11-16,96,7.DOI:10.3969/j.issn.1009-9492.2025.11.003
复杂驾驶场景下三维高斯溅射重建方法的性能优化
Performance Optimization of 3D Gaussian Splatting Reconstruction Method for Complex Driving Scenarios
摘要
Abstract
3D reconstruction in complex driving scenes is often affected by issues such as motion blur,loss of image details,and limited adaptability to lighting variations.To improve the reconstruction performance of 3D Gaussian Splatting(3DGS)in such scenarios,this study proposes an optimization method.First,a temporal consistency regularization mechanism is introduced to analyze motion smoothness between adjacent frames,ensuring temporal stability of the reconstruction results.Second,an edge enhancement strategy is applied to improve the clarity and sharpness of image edges.Finally,a loss function is constructed and the gradient update process is optimized to iteratively refine model parameters and generate high-quality rendered images.Experimental results demonstrate that,compared with existing methods,the proposed approach achieves average improvements of 8.7%and 4.2%in Peak Signal-to-Noise Ratio(PSNR)and Structural Similarity Index(SSIM),respectively,and reduces the Learned Perceptual Image Patch Similarity(LPIPS)by 7.4%.This study provides a novel optimization perspective for dynamic scene reconstruction in fields such as autonomous driving and virtual reality.关键词
三维高斯溅射/场景重建/动态/驾驶场景Key words
3D gaussian splatting/scene reconstruction/dynamic/driving scenarios分类
信息技术与安全科学引用本文复制引用
朱伟枝,谢娟烘..复杂驾驶场景下三维高斯溅射重建方法的性能优化[J].机电工程技术,2025,54(11):11-16,96,7.基金项目
广东省青年创新人才类项目(2024KQNCX118) （2024KQNCX118）
广东理工学院科技项目(2024ZDZK004) （2024ZDZK004）
