MQDF with a novel covariance matrix estimation and discriminant LSRC, which is better for in-air handwritten Chinese character recognition.
Zhaochun XuWeiqiang WangKe LuPublished in: ICIP (2016)
Keyphrases
- covariance matrix
- handwritten chinese character recognition
- principal component analysis
- estimation error
- covariance matrices
- positive definite
- sample size
- gaussian mixture
- objective function
- mahalanobis distance
- pseudo inverse
- eigendecomposition
- geometrical interpretation
- feature extraction
- correlation matrix
- multivariate gaussian
- face recognition
- computer vision
- class conditional densities
- cma es
- eigenvalues and eigenvectors
- machine learning
- special case
- dimensionality reduction
- optimization algorithm
- low dimensional