An RGB-D salient object detection method based on visual liquid time constant and cross-modal wavelet fusion

By fusing visual liquid time constant with cross-modal wavelet, the structural destruction and high-frequency detail degradation problems of the Mamba architecture in the global modeling process are solved, achieving high-precision salient target detection and cross-scene robustness.

CN122289884APending Publication Date: 2026-06-26KUNMING UNIV OF SCI & TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202610177247.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-02-06
Publication Date
2026-06-26

AI Technical Summary

Technical Problem

The existing Mamba architecture causes structural damage and high-frequency detail degradation during global modeling, and the lack of explicit decoupling and fusion between RGB and depth modes limits cross-modal collaborative effects.

Method used

A method combining visual liquid time constant and cross-modal wavelet fusion is adopted. The visual liquid time constant module realizes dynamic spatiotemporal gating, and the cross-modal wavelet fusion module performs frequency decoupling and complementary fusion to improve the model's ability to depict object boundaries and texture details.

Benefits of technology

It effectively solves the problems of structural destruction and high-frequency detail degradation in the global modeling process of the Mamba architecture, achieves high-precision salient target detection, and demonstrates excellent cross-modal collaborative efficiency and cross-scene generalization robustness.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122289884A_ABST
    Figure CN122289884A_ABST
Patent Text Reader

Abstract

This invention relates to an RGB-D salient target detection method based on visual liquid time constant and cross-modal wavelet fusion, belonging to the field of image processing technology. The method includes: using cascaded visual Mamba blocks to perform feature encoding on the acquired preprocessed RGB image and corresponding depth map, extracting multi-scale RGB features and multi-scale depth features; inputting the multi-scale RGB features and multi-scale depth features into a constructed visual liquid time constant module to obtain enhanced multi-scale RGB features and enhanced multi-scale depth features; inputting the enhanced multi-scale RGB features and enhanced multi-scale depth features into a constructed cross-modal wavelet fusion module to obtain cross-modal fused features; inputting the cross-modal fused features into a cascaded multi-scale decoder, and outputting an RGB-D salient target detection prediction map by constructing a total loss function. This aims to solve the technical problems of structural destruction and high-frequency detail degradation caused by existing Mamba architectures during global modeling.
Need to check novelty before this filing date? Find Prior Art