Abstract
In this paper, we proposed a hardware system with Field Programmable Gate Array (FPGA) for facial expression recognition which used Haar Discrete Wavelet Transform (DWT) and Cerebellar Model Articulation Controller (CMAC). Firstly, the facial expression features are automatically extracted and preprocessed to obtain the frontal view of faces. A 2D DWT IP is then used to decrease the size of images. Thirdly, a block size of the lower frequency of DWT coefficients is rearranged as input vectors with binary manner to send into the proposed CMAC IP that can rapidly obtain output using non-linear mapping with look-up table in training or recognizing phase. Finally, the experimental results demonstrated recognition rates with a block size of coefficient in lower frequency to recognize six expressions, including happiness, sadness, surprise, anger, disgust and natural to show promising recognition results.