Remote sensing image neural rendering method, electronic equipment and storage medium

By using attention-aware multi-scale feature extraction and neural coding feature volume construction, the problems of high computational resource consumption and insufficient accuracy in traditional remote sensing image 3D reconstruction are solved, achieving efficient and accurate remote sensing image rendering.

CN121033247AActive Publication Date: 2025-11-28AEROSPACE INFORMATION RES INST CAS
7 Cites 0 Cited by

Patent Information

Application Number
CN202511560371.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-10-29
Publication Date
2025-11-28
Estimated Expiration
2045-10-29

AI Technical Summary

Technical Problem

Traditional neural radiation field methods consume large amounts of computational resources and have slow training speeds in 3D reconstruction of remote sensing images. They are difficult to process high-resolution images and lack effective geometric constraints and sufficient surface feature extraction, resulting in insufficient reconstruction accuracy. MVS-based methods require accurate depth maps, which are difficult to obtain, and cannot process images lacking texture or non-Lambertian surfaces.

Method used

We employ attention-aware multi-scale feature extraction, which obtains target feature maps of different scales through staged processing. We then combine rational polynomial deformation and 3D convolution processing to construct a neural coding feature volume and perform shadow radiation processing to generate rendering results of remote sensing images.

Benefits of technology

It improves the rendering quality and efficiency of remote sensing images, enables training without 3D supervision, enhances the focus on key areas and information capture, and improves the accuracy of detailed feature representation and scene reconstruction.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention provides a remote sensing image neural rendering method, electronic equipment and a storage medium, which can be applied to the technical field of remote sensing image rendering. The method comprises the following steps: performing feature extraction of attention perception on a remote sensing image in stages to obtain a plurality of target feature maps with different scales; performing rational polynomial deformation on the plurality of target feature maps with different scales in stages to obtain a plurality of single-scale cost volumes, and aggregating the plurality of single-scale cost volumes into a feature map cost volume; performing 3D convolution processing on the feature map cost volume to obtain a neural coding feature volume, and performing shadow radiation processing on the neural coding feature volume and a light sampling point of the remote sensing image to obtain a regression rendering attribute parameter of the remote sensing image; and performing image neural rendering on the remote sensing image by using the regression rendering attribute parameter to obtain a rendering result of the remote sensing image.
Need to check novelty before this filing date? Find Prior Art