Estimation of Mueller matrices using non-local means filtering.

Area of Science:

• Optics and Photonics
• Image Processing
• Polarimetry

Background:

• The Mueller imaging framework offers a comprehensive description of polarization properties.
• Previous work focused on the simpler Stokes imaging framework.
• Estimating Mueller matrix images is complex due to higher dimensionality and potential for non-physical results.

Purpose of the Study:

• To extend non-local means filtering for estimating polarization signatures in the Mueller imaging framework.
• To develop a physically admissible estimation procedure for Mueller matrices.
• To enable noise reduction and edge preservation in Mueller matrix images.

Main Methods:

• Application of non-local means filtering to Mueller matrix data.
• Utilizing Cholesky decomposition of the coherency matrix for physical admissibility.
• Development of a novel parameterization for the Mueller matrix boundary set.

Main Results:

• Successful noise removal from Mueller matrix images.
• Preservation of important image features and edges.
• Demonstration of the method's effectiveness on both synthetic and real-world data.

Conclusions:

• The proposed unsupervised method effectively estimates physically admissible Mueller matrices.
• This approach advances polarization imaging by extending filtering techniques to a more complex framework.
• The method shows promise for various applications in optical imaging and material characterization.