Face recognition using improved local texture patterns
Wankou Yang, Changyin Sun · 2011
Recently, local texture patterns based methods have been widely used in face recognition. LBP and LTP are two typical feature descriptor methods. LBP chooses the central pixel as the threshold and the central pixel is easy be noised. In LTP, it is difficult to automatically set a suitable threshold to overcome the noise. As we know, the average and the stand deviation of a region are good represent of the region and robust to noise. So we present an improved LBP method by replacing the central pixel with the average of the region, an improved LTP method by replacing the central pixel with the average of the region. The experimental results on ORL, AR face databases show that our present methods have better performance than LBP and LTP.