Linear discriminant analysis (LDA) is a widely used method for supervised feature extraction and dimension reduction in pattern recognition and data analysis. However, the existing data is characterized by high dimensionality and complexity, often containing numerous redundant features. These interference features may misguide the discriminative capabilities of traditional LDA. Therefore, how to extract effective discriminative features from plenty of information is the challenge to be addressed. To solve above problems, we propose a novel method called Enhanced Discriminant Sparse Feature Extraction (EDSFE). Specifically, EDSFE differs from traditional LDA methods through the incorporation of novel components that enhance its discriminative feature extraction capabilities. While both EDSFE and traditional LDA aim to maximize the between-class difference and minimize the within-class difference, EDSFE introduces unique innovations to improve its performance. It promotes sparsity and reconstruction capability of the projection matrix through sparse regularization and reconstruction error terms. Extensive experiments conducted on three databases shows that the proposed method performs competitively when compared to other state-of-the-art feature extraction methods.