A Comprehensive Review of Vision-Based 3D Reconstruction Methods

Sensors, Journal Year: 2024, Volume and Issue: 24(7), P. 2314 - 2314

Published: April 5, 2024

With the rapid development of 3D reconstruction, especially emergence algorithms such as NeRF and 3DGS, reconstruction has become a popular research topic in recent years. technology provides crucial support for training extensive computer vision models advancing general artificial intelligence. deep learning GPU technology, demand high-precision high-efficiency information is increasing, fields unmanned systems, human-computer interaction, virtual reality, medicine. The becoming inevitable. This survey categorizes various methods technologies used reconstruction. It explores classifies them based on three aspects: traditional static, dynamic, machine learning. Furthermore, it compares discusses these methods. At end survey, which includes detailed analysis trends challenges development, we aim to provide comprehensive introduction individuals who are currently engaged or planning conduct Our goal help gain understanding relevant knowledge related

Language: Английский

---

SACNet: A Novel Self-Supervised Learning Method for Shadow Detection from High-Resolution Remote Sensing Images

Journal of Geovisualization and Spatial Analysis, Journal Year: 2025, Volume and Issue: 9(1)

Published: Feb. 24, 2025

Language: Английский

---

Multi-angle analysis of shading-based modelling: extracting the urban building height based on ZY-3 three-line-array camera

International Journal of Remote Sensing, Journal Year: 2025, Volume and Issue: unknown, P. 1 - 33

Published: May 7, 2025

Language: Английский

---

Research on Defect Detection Method of Fusion Reactor Vacuum Chamber Based on Photometric Stereo Vision

Sensors, Journal Year: 2024, Volume and Issue: 24(19), P. 6227 - 6227

Published: Sept. 26, 2024

This paper addresses image enhancement and 3D reconstruction techniques for dim scenes inside the vacuum chamber of a nuclear fusion reactor. First, an improved multi-scale Retinex low-light algorithm with adaptive weights is designed. It can recover detail information that not visible in environments, maintaining clarity contrast easy observation. Second, according to actual needs target plate defect detection chamber, based on photometric stereo vision proposed. To optimize position light source, source illumination profile simulation system designed this provide optimized array crack chambers without need extensive experimental testing. Finally, robotic platform mounted binocular stereo-vision camera constructed experiments are performed separately. The results show above method broaden gray level low-illumination images improve brightness value contrast. maximum depth error less than 24.0% width 15.3%, which achieves goal detecting reconstructing defects chamber.

Language: Английский

---

Building Height Extraction Based on Spatial Clustering and a Random Forest Model

ISPRS International Journal of Geo-Information, Journal Year: 2024, Volume and Issue: 13(8), P. 265 - 265

Published: July 26, 2024

Building height (BH) estimation is crucial for urban spatial planning and development. BH using digital surface model data involves obtaining ground roof elevations. However, vegetation shadows around buildings affect the selection of required elevation, resulting in large errors. In highly urbanized areas, similar heights often have characteristics proximity, which reference significance but are rarely utilized. Herein, we propose a method based on BIRCH clustering random forest (RF) model. We obtain initial results optimal search area multi-index evaluation. an RF classification used to match their distance attribute characteristics. Finally, adjusted elevation obtained from secondary screening matching. The validation two areas with over 12,000 show that proposed reduces root-mean-square error final compared results. Comparing maps shows produce relatively accurate high shading coverage, as well dense buildings. Thus, has been validated its effectiveness reliability.

Language: Английский

---

SingleRecon: Reconstructing Building 3D models of LoD1 from A Single Off-Nadir Remote Sensing Image

IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, Journal Year: 2024, Volume and Issue: 17, P. 19588 - 19600

Published: Jan. 1, 2024

Language: Английский

---