Regularizing covariance estimation by quantized eigenvalues and its application to image classification

Regularizing sample covariance estimation by quantized eigenvalues is proposed to mitigate the bias problem of the sample covariance estimation. Quantized eigenvalues are used to regularize the sample covariance estimation. In this sense, the regularized discriminant analysis can be considered as a special case of one quantization level, which is the average of eigenvalues. The proposed algorithm is applied to image classification and experimental results show that the proposed algorithm improves the classification performance.