Multi-View Three-Dimensional Reconstruction Based on Feature Enhancement and Weight Optimization Network
Bibliographic Data
| ID | 22032124 |
|---|---|
| Authors | Guobiao Yao (0000-0001-6305-4514, Shandong Jianzhu University, corresponding author), Ziheng Wang (0000-0002-3953-0493, Shandong Jianzhu University), Guozhong Wei (Shandong Provincial Communications Planning and Design Institute (China)), Fengqi Zhu (Shandong Provincial Communications Planning and Design Institute (China)), Qingqing Fu (0000-0002-1489-8865, Shandong Jianzhu University), Qian Yu (0000-0003-3612-6951, Shandong Provincial Communications Planning and Design Institute (China)), Min Wei (0000-0002-3131-0955, State Key Laboratory of Remote Sensing Science) |
| Year | 2025 |
| Volume | 14 |
| Issue | 2 |
| Pages | 43 |
| Publication date | 2025-01-24 |
| Open Access | Yes |
| Type | ARTICLE |
| Venue | ISPRS International Journal of Geo-Information (JOURNAL) |
| Journal identifiers | ISSN: 2220-9964 • E-ISSN: 2220-9964 |
| Publisher | MDPI AG (PUBLISHER • IT) |
| DOI | 10.3390/ijgi14020043 |
| OpenAlex | W4406806057 |
| Language | EN |
| References cited | 23 |
Aiming to address the issue that existing multi-view stereo reconstruction methods have insufficient adaptability to the repetitive and weak textures in multi-view images, this paper proposes a three-dimensional (3D) reconstruction algorithm based on Feature Enhancement and Weight Optimization MVSNet (Abbreviated as FEWO-MVSNet). To obtain accurate and detailed global and local features, we first develop an adaptive feature enhancement approach to obtain multi-scale information from the images. Second, we introduce an attention mechanism and a spatial feature capture module to enable high-sensitivity detection for weak texture features. Third, based on the 3D convolutional neural network, the fine depth map for multi-view images can be predicted and the complete 3D model is subsequently reconstructed. Last, we evaluated the proposed FEWO-MVSNet through training and testing on the DTU, BlendedMVS, and Tanks and Temples datasets. The results demonstrate significant superiorities of our method for 3D reconstruction from multi-view images, with our method ranking first in accuracy and second in completeness when compared to the existing representative methods
Advanced Vision and Imaging · Computer Science · Image Processing Techniques and Applications · Optical measurement and interference techniques · Artificial Intelligence
| Highly cited | No |
|---|