Resolution enhancement via guided filtering for spatial-frequency multiplexing single-shot high-speed imaging

Area of Science:

• Optics and Photonics
• Image Processing
• High-Speed Imaging

Background:

• The Frequency Recognition Algorithm for Multiple Exposures (FRAME) technique offers high temporal and spatial resolution in single-shot high-speed videography.
• FRAME suffers from crosstalk, leading to artifacts and reduced resolution in recovered frames.

Purpose of the Study:

• To develop an improved FRAME reconstruction method for minimizing artifacts and enhancing spatial resolution.
• To address the limitations of existing FRAME techniques through advanced image processing.

Main Methods:

• Proposed a FRAME reconstruction method incorporating guided filters for zero-frequency component removal.
• Implemented total variation (TV) denoising to further eliminate artifacts and optimize performance.
• Validated the method through simulations and experimental setups, including a two-frame system.

Main Results:

• Achieved a 1.8-fold improvement in spatial resolution on a USAF 1951 target, increasing from 16 lp/mm to 28.5 lp/mm.
• Successfully resolved two fuel spray droplets separated by 30 μm in high-speed dynamic scenarios.
• Demonstrated superior performance compared to traditional FRAME methods in complex imaging conditions.

Conclusions:

• The proposed guided filter and TV denoising method effectively removes artifacts and enhances spatial resolution in FRAME.
• This advanced post-processing technique significantly improves the quality and detail of high-speed videography.
• The method shows robust and efficient capabilities for analyzing fast dynamic phenomena.