Focused image recovery from two defocused images recorded with different camera settings

Area of Science:

• Computer Vision
• Image Processing
• Computational Imaging

Background:

• Defocused images are common in photography due to variations in camera parameters.
• Recovering sharp images from blurred data is crucial for various applications.

Purpose of the Study:

• To introduce two novel methods for focused image recovery from two defocused images.
• To compare the computational efficiency and performance of the proposed methods.

Main Methods:

• Estimating blur parameter sigma using depth-from-defocus.
• Spatial domain deconvolution requiring only sigma of the point spread function (PSF).
• Fourier domain deconvolution using a Wiener filter, requiring full PSF knowledge and a calibration procedure.

Main Results:

• Both methods successfully recovered focused images from experimental data.
• The spatial domain method is computationally less intensive.
• The spatial domain method performs satisfactorily for medium blur, while the Fourier domain method handles higher blur levels.

Conclusions:

• The proposed methods offer effective solutions for focused image recovery.
• The choice of method depends on the required accuracy and computational resources.
• Further research can explore optimizing these deconvolution techniques.