Face recognition based on Log-Gabor filter binary transformation

In this paper, a novel face recognition model based on the principle of high dimensional biomimetic information theory which is based on “things cognition” instead of “things classification”, and is more close to the human beings behavior is proposed. By Log-Gabor filtering, 24 Gabor sub-images were gained, then by thresholding these sub-images for binary transformation. Decision is made by minimum Hamming distances. Experiments performed on the JDL_A databases show the provided method can achieve a better performance than the baseline PCA method.