Windowing Approach for Face Recognition Using the Spatial-Temporal Method and Artificial Neural Network

Face recognition (FR) is getting a lot of attention for a good reason in the field of research and making a big impact in areas such as computer vision and human-computer interaction. This paper proposes a FR model based on the windowing technique using discrete cosine transform (DCT), average covariance and artificial neural network (ANN). The windowing technique is used to divide the whole image into 4 × 4, 8 × 8 and 16 × 16 size of windows. The DCT is applied to each window to acquire DCT coefficients. The average covariance is calculated for the obtained DCT coefficient matrix. The calculation of an average covariance decreases the original size of the image by around 97%. The network is created, trained and tested to assess the performance of the network using nine standard face databases. Experimental results indicate that the proposed model achieves a higher recognition rate with a reduced number of features and computational intricacy compared with conventional methods