Maxar International Sweden AB (20240233262). METHOD FOR 3D RECONSTRUCTION FROM SATELLITE IMAGERY simplified abstract
Contents
METHOD FOR 3D RECONSTRUCTION FROM SATELLITE IMAGERY
Organization Name
Inventor(s)
[[:Category:Tim Yngesj� of Linköping (SE)|Tim Yngesj� of Linköping (SE)]][[Category:Tim Yngesj� of Linköping (SE)]]
Carl Sundelius of Linköping (SE)
Anton Nordmark of Linköping (SE)
METHOD FOR 3D RECONSTRUCTION FROM SATELLITE IMAGERY - A simplified explanation of the abstract
This abstract first appeared for US patent application 20240233262 titled 'METHOD FOR 3D RECONSTRUCTION FROM SATELLITE IMAGERY
The present disclosure involves a method for 3D reconstruction from satellite imagery using deep learning. This method includes providing at least two overlapping 2D satellite images, imaging device parameters for the images, and at least one trained machine learning network (MLN) capable of predicting depth maps.
- The method involves using overlapping 2D satellite images and imaging device parameters.
- A trained MLN is utilized to predict depth maps based on multi-view geocoded 3D ground truth data.
- The MLN is trained on a training set to accurately predict depth maps of the provided satellite images.
Potential Applications: - Urban planning and development - Environmental monitoring - Disaster response and recovery efforts
Problems Solved: - Accurate 3D reconstruction from satellite imagery - Efficient depth map prediction using deep learning
Benefits: - Enhanced spatial understanding from satellite images - Improved decision-making in various industries - Cost-effective and time-efficient 3D reconstruction process
Commercial Applications: Title: Advanced 3D Reconstruction Technology for Satellite Imagery Analysis This technology can be used in industries such as urban planning, agriculture, and forestry for precise mapping and analysis of landscapes.
Prior Art: Researchers can explore existing patents related to satellite image analysis, deep learning, and 3D reconstruction techniques to understand the advancements in this field.
Frequently Updated Research: Stay updated on advancements in deep learning algorithms for satellite image analysis and 3D reconstruction techniques to enhance the accuracy and efficiency of the process.
Questions about 3D Reconstruction from Satellite Imagery using Deep Learning: 1. How does deep learning improve the accuracy of depth map prediction from satellite images? 2. What are the key challenges in training machine learning networks for 3D reconstruction from satellite imagery?
Original Abstract Submitted
the present disclosure relates to a method for 3d reconstruction from satellite imagery using deep learning, said method comprising providing () at least two overlapping 2d satellite images, providing () imaging device parameters for the at least two overlapping 2d satellite images, providing () at least one trained machine learning network, mln, able to predict depth maps, said trained mln being trained on a training set comprising multi-view geocoded 3d ground truth data and predicting () a depth map of the at provided at least two 2d satellite images using the trained at least one mln and based on the corresponding imaging device parameters.
