Automatic detection of feather defects using Lie group and fuzzy Fisher criterion for shuttlecock production

Abstract

Feathers are the main raw material of shuttlecock. The quality of feather is a key factor in shuttlecock production and detection of feather defects is a challenging problem. In this paper, an automatic detection for feather defects was proposed that relies on fuzzy, Lie Group and machine learning theory. First, the feather images were enhanced using fuzzy set. Then, a covariance matrix was constructed as features of the defect region and Riemannian metric was assigned on a Riemannian manifold with a Lie group structure. The mean covariance matrix with nonsymmetric structure was proved to have a Lie group structure. To establish the automatic detection of feather defects, this study introduced Riemann mean and membership degree into the machine learning method of Fisher linear discriminant to implement feature recognition. Experiments were performed to demonstrate the feasibility of features of defect region as a differential manifold.