Edge-Aware Unidirectional Total Variation Model for Stripe Non-Uniformity Correction.

Area of Science:

• Infrared imaging systems
• Image processing
• Computer vision

Background:

• Stripe non-uniformity is a common noise issue in array-based infrared imaging.
• Total variation models reduce noise but can over-smooth images, losing details.
• Existing de-striping techniques struggle with preserving image structure.

Purpose of the Study:

• To develop an advanced de-striping algorithm for infrared images.
• To address the over-smoothing limitations of traditional methods.
• To enhance edge and texture preservation during noise correction.

Main Methods:

• A novel correction scheme based on a unidirectional variation model.
• Incorporation of edge-aware weighting to retain image structure.
• Introduction of statistical-based regularization for improved edge performance.

Main Results:

• The proposed method significantly reduces stripe non-uniformity.
• Demonstrated superior edge preservation compared to state-of-the-art techniques.
• Achieved better overall image correction performance on real-world data.

Conclusions:

• The unidirectional variation model with edge-aware weighting and statistical regularization offers an effective de-striping solution.
• This approach overcomes the over-smoothing issue, preserving critical image details.
• The method shows significant promise for infrared imaging applications.