Separation of superimposed images with subpixel shift.

Area of Science:

• Image processing
• Computational imaging
• Signal separation

Background:

• Superimposed images present challenges in analysis, especially with moving foreground patterns.
• Separating distinct foreground patterns from a steady background requires specialized algorithms.
• Artifacts in imaging, such as dust or scratches, can obscure important data.

Purpose of the Study:

• To develop an original algorithm for separating superimposed images.
• To address the specific case of a steady background and independently moving foreground patterns.
• To provide a method for cleaning X-ray tomographic radiographs contaminated by moving artifacts.

Main Methods:

• A novel algorithm is proposed for image separation.
• The method assumes a single pair of superimposed images with sub-pixel displacements.
• It leverages a smoothly varying background and compact foreground supports.

Main Results:

• The algorithm effectively separates superimposed images with moving foreground patterns.
• Demonstrated performance on X-ray tomographic radiographs with dust particles and surface scratches.
• An automatic and simple treatment was developed to remove feature effects.

Conclusions:

• The proposed algorithm offers an effective solution for superimposed image separation.
• It successfully removes artifacts in X-ray imaging caused by moving particles or surface defects.
• The method provides a practical approach for image cleaning and artifact removal.