Physical model guidance and Mangban combined cloud removal method

Through the combination of physical model guidance and Mamba, the cloud removal problem in the coexistence environment of thin clouds and thick clouds in optical remote sensing images is solved, and the efficient cloud removal effect is achieved, ensuring the accuracy and detail fidelity of cloud removal results. .

CN118967527AActive Publication Date: 2024-11-15ANHUI UNIV

Patent Information

Application Number
CN202411443444.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-10-16
Publication Date
2024-11-15
Estimated Expiration
2044-10-16

AI Technical Summary

Technical Problem

Existing optical remote sensing image cloud removal methods are difficult to achieve efficient cloud removal in complex environments where thin clouds and thick clouds coexist, resulting in excessive correction of thin cloud areas and insufficient filling of thick cloud areas in the reconstruction results.

Method used

Using a cloud removal method combined with physical model guidance and Mamba, the cloud region is divided by combining the atmospheric radiation transmission model by constructing data sets, estimating the cloud distribution significance graphs guided by physical model, and building and training of the Mamba-based cloud removal model. , using the Mamba visual state space basic module for cloud removal, and the adaptive area aggregation module integrates the removal results.

Benefits of technology

The cloud removal performance is significantly improved, ensuring that the cloud removal results are not overly corrected in the thin cloud area and the thick cloud area is fully filled to obtain cloud removal results with color fidelity, clear details and high contrast.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118967527A_ABST
    Figure CN118967527A_ABST
Patent Text Reader

Abstract

The invention relates to a physical model guidance and Mangban combined cloud removal method. The method comprises the following steps: constructing an optical remote sensing cloud removal data set; carrying out cloud distribution saliency map region division based on physical model guidance; building a cloud removal model based on Mangbar and training model parameters; obtaining cloud optical remote sensing image data to be processed; and verifying the cloud removal performance. Compared with the prior art, the degree of cloud interference of the cloud coverage remote sensing image is calculated and analyzed in an auxiliary manner by using the saliency map, and the cloud removal model is guided to carry out targeted parameter adjustment so as to enhance the sensitivity of the model to information recovery in a specific area. And meanwhile, a cloud removal model is constructed based on a Mangban state space model with efficient global feature extraction capability, so that the recovery of missing information in thin cloud and thick cloud regions is realized. Under the condition that thin cloud and thick cloud coexist, it is guaranteed that the cloud removal result is not excessively corrected in the thin cloud area, and the thick cloud area is fully filled and optimized.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Optical remote sensing image cloud and mist removing method and system based on generative adversarial network

    CN116563728A

  • Semantic header detection using pre-trained embeddings

    US20210109993A1

  • Method for removing cloud and haze on basis of depth channel sensing

    WO2022095253A1

Cited By

  • Seismic data interpolation method based on Mama

    CN119622231A

  • Quality-graded remote sensing image composite cloud removal method, system and equipment

    CN119671898A

  • Method and system for calculating vegetation coverage index based on dual dynamic cloud optimization mechanism

    CN120849877A

  • Method and system for calculating vegetation cover index based on double dynamic cloud optimization mechanism

    CN120849877B

  • Physical guidance-based multispectral remote sensing image progressive diffusion cloud removal method

    CN122434790A