A multi-scale lining image stitching method and computer program product for tunnel detection parallax scenarios.

CN122089564AActive Publication Date: 2026-05-26UESTC (SHENZHEN) ADVANCED RES INST +2
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
UESTC (SHENZHEN) ADVANCED RES INST
Filing Date
2026-04-23
Publication Date
2026-05-26

AI Technical Summary

Technical Problem

Existing methods for stitching lining images in parallax scenarios during tunnel inspection suffer from low multi-scale feature alignment accuracy, insufficient non-rigid registration, difficulty in completely eliminating stitching artifacts, poor adaptability, and difficulty in ensuring the quality of lining image stitching.

Method used

Multi-scale feature extraction is performed using a feature pyramid network based on a convolutional neural network. Image registration is performed by combining global homography matrix iterative optimization and TPS interpolation algorithm. High-quality panoramic images are generated through masking and artifact hiding methods.

Benefits of technology

It achieves high-precision image stitching in tunnel inspection parallax scenarios, adapts to scenarios with different parallax levels, and outputs wide-view, high-resolution naturally stitched panoramic images, improving the stitching quality of lining images.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122089564A_ABST
    Figure CN122089564A_ABST
Patent Text Reader

Abstract

This invention discloses a multi-scale lining image stitching method and computer program product for tunnel detection parallax scenarios, relating to the field of computer vision technology. It solves the technical problem of poor adaptability and difficulty in ensuring stitching quality in traditional lining image stitching methods. The method includes: performing feature map matching on a first image and a second image to obtain an initial global homography matrix; performing multi-scale matrix iterative optimization and generating a global aligned feature map to obtain an optimal global homography matrix; performing local non-rigid adjustment on the second image using a TPS interpolation algorithm; extracting initial masks from the registered first and second images and generating seam masks; finding the optimal seam within the seam masks to generate a content mask; and fusing the registered first and second images based on the content mask to output a panoramic image. This invention exhibits good adaptability to scenarios with varying degrees of parallax, significantly improving the stitching quality of lining images.
Need to check novelty before this filing date? Find Prior Art