Image processing method and device, electronic equipment, storage medium and program product

By using image processing methods and image acquisition devices to acquire multiple frames of scene images, feature points are extracted for pose estimation and dense depth estimation, which solves the problems of high cost and large size of depth sensors and realizes efficient scene reconstruction of lightweight devices.

CN122244285APending Publication Date: 2026-06-19GRAVITYXR ELECTRONICS & TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
GRAVITYXR ELECTRONICS & TECH CO LTD
Filing Date
2024-12-17
Publication Date
2026-06-19

AI Technical Summary

Technical Problem

Existing depth sensors are expensive, bulky, and complex to install in applications such as robot navigation and autonomous driving, which limits their use in lightweight devices.

Method used

By using image processing methods, multiple frames of scene images are acquired using image acquisition equipment, image feature points are extracted, pose estimation is performed, a sparse point cloud is established and a sparse depth map is projected, and a dense depth map is constructed by combining a dense depth estimation model to achieve scene reconstruction.

Benefits of technology

It can acquire depth information of each spatial feature point in a scene without relying on a depth sensor, simplifying the device structure, improving the accuracy of scene reconstruction, and enhancing the portability and scalability of terminal devices.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122244285A_ABST
    Figure CN122244285A_ABST
Patent Text Reader

Abstract

This application provides an image processing method, apparatus, electronic device, storage medium, and program product. The method includes: acquiring multiple frames of scene images of a target scene; performing pose estimation on an image acquisition device based on image feature points contained in the multiple frames of scene images to obtain pose information of the image acquisition device when acquiring the multiple frames of scene images; stitching together spatial feature points corresponding to the image feature points based on the pose information to establish a sparse point cloud; projecting the sparse point cloud to obtain a sparse depth map corresponding to the pose information of the current frame; obtaining a dense depth map based on the sparse depth map and the image information corresponding to the scene images; and reconstructing the scene based on the dense depth map and real-time pose information to obtain a reconstructed scene corresponding to the real-time pose. This method aims to achieve fast and accurate 3D reconstruction.
Need to check novelty before this filing date? Find Prior Art