Denoising multicriterion iterative reconstruction in emission spectral tomography

Area of Science:

• Optical testing and fluid dynamics analysis.
• Three-dimensional reconstruction of physical parameters in fluid fields like flames and plasmas.

Background:

• Computed tomography is crucial for reconstructing 3D fluid field distributions.
• Projection data in optical testing are frequently corrupted by noise, impacting reconstruction accuracy.

Purpose of the Study:

• To enhance the performance of optical computed tomography in reconstructing noisy projection data.
• To develop and evaluate a novel algorithm combining denoising and iterative reconstruction for improved accuracy.

Main Methods:

• A self-adaptive prefiltering denoising approach (SPDA) was developed to estimate noise levels using frequency domain statistics.
• A Butterworth low-pass filter's cutoff frequency was determined by noise energy.
• The SPDA was integrated with a multicriterion iterative reconstruction (MCIR) algorithm for limited-view reconstruction.

Main Results:

• Simulated reconstructions of test phantoms demonstrated the algorithm's effectiveness.
• A flame emission spectral tomography experiment validated the SPDA-MCIR performance on real-world data.
• The SPDA-MCIR combination showed marked improvements in reconstructing noisy optical computed tomography data compared to traditional methods.

Conclusions:

• The proposed SPDA-MCIR algorithm offers a significant advancement for optical computed tomography in noisy environments.
• This approach enhances the reliability and accuracy of 3D fluid field reconstructions from corrupted projection data.