Iterative Blind Deconvolution Algorithm for Deblurring a Single PSP/TSP Image of Rotating Surfaces

Area of Science:

• Fluid dynamics
• Optical measurement techniques
• Image processing

Background:

• Imaging pressure-sensitive paint (PSP) on moving surfaces suffers from motion blur due to finite exposure times.
• High-sensitivity PSP with long luminescence decay times exacerbate image blurring, particularly at blade edges.
• Current deconvolution methods often assume a spatially invariant time constant, leading to inaccuracies with strong pressure gradients.

Purpose of the Study:

• To develop an iterative image deconvolution method for accurate PSP imaging on moving surfaces.
• To address the limitations of spatially invariant point spread functions (PSFs) in PSP deblurring.
• To implement a spatially variant PSF that accounts for local pressure variations.

Main Methods:

• Introduced an iterative deconvolution scheme employing a spatially variant PSF.
• Estimated point-by-point PSF values iteratively, as the time constant depends on local pressure.
• Utilized a kernel-free forward model for generating test images and a spinning disk setup for real-world validation.

Main Results:

• The iterative method successfully estimated a super-resolved, spatially varying blur kernel with sub-pixel resolution.
• Restored images using classical iterative regularization tools showed improved accuracy.
• Demonstrated applicability across various experimental conditions, including strong pressure gradients.

Conclusions:

• The proposed iterative deconvolution with a spatially variant PSF effectively reduces motion blur in PSP imaging.
• This method overcomes the inaccuracies of previous spatially invariant approaches, especially in areas with significant pressure changes.
• The technique offers enhanced accuracy for pressure measurements on dynamic surfaces.