.png){kind=icon}
.png){kind=icon}
Return
Efficient multi-objective dynamic programming for building mesh optimization with 3D lines
.png){kind=icon}
PRE
AI
DOI：10.1016/j.isprsjprs.2026.02.033.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
3D meshes with accurate edges are crucial for visualization and structural understanding of buildings. Currently, the construction of 3D meshes is mainly based on 3D points. Although 3D lines can provide effective constraints on edges and planes, they have received limited attention in existing research. This paper presents a novel dynamic programming (DP) method to efficiently optimize the existing building mesh with 3D lines, which ensures local topological consistency and eliminates the need for global retriangulation. For accurate retriangulation, we propose an adaptive weighting strategy to optimize multiple objectives during DP. For efficient retriangulation, we introduce a probabilistic decision model to reduce the redundant optimization process in DP. Comparative experiments demonstrated that (1) triangulation with DP achieved optimal geometric accuracy and is robust to noise with our adaptive weighting strategy; and (2) our probabilistic decision model reduced the running time of DP by approximately 25%, with little impact on accuracy. The source code of our work is available at https://github.com/XiaotongLi123/mesh-optimization-with-3D-building-lines .
Keywords:
3D lines
dynamic programming
mesh optimization
building reconstruction
multi-objective optimization
Journal
IF:
12.2
Papers:
4.3K
Citations:
3.2W