Scene-based nonuniformity correction and enhancement: pixel statistics and subpixel motion.

Area of Science:

• Image processing
• Sensor technology
• Computer vision

Background:

• Focal-plane array sensors often suffer from nonuniformity, degrading image quality.
• Existing methods for nonuniformity correction (NUC) may not fully address structured noise or enhance resolution.
• Image enhancement and resolution improvement are critical for many sensor applications.

Purpose of the Study:

• To propose a comprehensive framework for scene-based nonuniformity correction and enhancement (NUCE).
• To introduce a novel registration-based nonuniformity correction super-resolution (NUCSR) method.
• To achieve simultaneous removal of severe nonuniformity and resolution enhancement using subpixel motion.

Main Methods:

• Development of a comprehensive imaging model and parametric motion estimation.
• Bootstrapping a novel NUCSR method using statistical scene-based NUC techniques.
• Utilizing a parametric motion model to handle diverse motion scenarios and enable super-resolution.

Main Results:

• Demonstrated removal of severe and structured nonuniformity in sensor images.
• Achieved simultaneous image resolution improvement through super-resolution.
• Validated the effectiveness of the NUCSR method and the overall NUCE framework on real-world data.

Conclusions:

• The proposed NUCE framework effectively corrects and enhances images from focal-plane array sensors.
• The NUCSR method is efficient and capable of handling parametric motions for superior super-resolution.
• The framework offers a robust solution for obtaining clean and high-resolution images.