Destriping of Remote Sensing Images by an Optimized Variational Model

Area of Science:

• Remote Sensing
• Image Processing
• Computer Vision

Background:

• Stripe noise in satellite imagery degrades image quality and hinders applications.
• Existing destriping methods often cause loss of detail or introduce artifacts.

Purpose of the Study:

• To develop a novel unidirectional variational model for effective horizontal stripe noise removal.
• To improve upon existing methods by preserving image details and reducing artifacts.

Main Methods:

• A unidirectional variational model incorporating directional characteristics and structural sparsity of stripe noise.
• Introduction of the ℓp quasinorm for enhanced sparse constraint description.
• Application of the fast alternating direction method of multipliers (ADMM) for solving the non-convex model.

Main Results:

• The proposed model effectively removes horizontal stripe noise.
• Demonstrated superior performance in preserving fine image details compared to existing methods.
• Achieved excellent destriping effects with improved efficiency and robustness.

Conclusions:

• The new variational model offers a significant advancement in satellite image destriping.
• The method provides a robust and efficient solution for removing stripe noise while maintaining image integrity.