A cv-based sustainable interactive image segmentation method

By employing a CV-based sustainable interactive image segmentation method, utilizing the Chan-Vese segmentation model and multiple interactive adjustments, the problem of segmentation results being sensitive to initial interaction points and lacking generalization ability in existing technologies is solved, achieving stable and interpretable image segmentation results.

CN122115469APending Publication Date: 2026-05-29HARBIN INST OF TECH
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
HARBIN INST OF TECH
Filing Date
2026-01-15
Publication Date
2026-05-29

AI Technical Summary

Technical Problem

Existing image segmentation techniques cannot effectively utilize information from multiple interactions, have difficult labeling methods, and the final segmentation results are sensitive to the initial interaction points, with limited generalization ability, especially in complex scenarios.

Method used

A sustainable interactive image segmentation method based on CV is adopted. The segmentation region is determined by calculating the gray value of pixels and interactive points. The Chan-Vese segmentation model is used in combination with forced constraints, which allows for multiple interactive adjustments to the segmentation results. The velocity field and level set function are modified to optimize the segmentation.

Benefits of technology

It achieves stable segmentation results without the need for high-quality labeled data. The model is not sensitive to initial interaction points, has good generalization ability and interpretability, and can optimize the segmentation effect through multiple interactions.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122115469A_ABST
    Figure CN122115469A_ABST
Patent Text Reader

Abstract

The application provides a sustainable interactive image segmentation method based on CV and belongs to the technical field of image global segmentation. The method is used for solving the problems that the information of multiple interactions cannot be introduced in the existing image segmentation, the marking mode is difficult, and the final segmentation result is sensitive to the marking points. The method comprises the following steps: S1, inputting a picture and calculating the gray value of a pixel point; S2, performing first interaction and confirming a region to be segmented according to the input interaction point; S3, initializing a level set and a velocity field, performing discrete level evolution, and completing first segmentation; S4, judging whether to continue interaction, if not, outputting a segmentation result, if yes, modifying the level set and the velocity field, and continuing segmentation; and S5, repeating S4 until no segmentation is needed, and outputting a final segmentation result.
Need to check novelty before this filing date? Find Prior Art