Feature Extraction on Mueller Matrix Data for Detecting Nonporous Electrospun Fibers Based on Mutual Information.

Lu Si,Xiaopeng Li,Yuanhuan Zhu,Yong Sheng,Hui Ma
DOI: https://doi.org/10.1364/oe.389181
IF: 3.8
2020-01-01
Optics Express
Abstract:The surface morphology of electrospun fibers largely determines their application scenarios. Conventional scanning electron microscopy is usually used to observe the microstructure of polymer electrospun fibers, which is time consuming and will cause damage to the samples. In this paper, we use backscattering Mueller polarimetry to classify the microstructural features of materials by statistical learning methods. Before feeding the Mueller matrix (MM) data into the classifier, we use a two-stage feature extraction method to find out representative polarization parameters. First, we filter out the irrelevant MM elements according to their characteristic powers measured by mutual information. Then we use Correlation Explanation (CorEx) method to group interdependent elements and extract parameters that represent their relationships in each group. The extracted parameters are evaluated by the random forest classifier in a wrapper forward feature selection way and the results show the effectiveness in classification performance, which also shows the possibility to detect nonporous electrospun fibers automatically in real time.
What problem does this paper attempt to address?