Median Ternary Pattern (MTP) for Face Recognition

This paper presents an effective appearance-based facial feature descriptor constructed with a new local texture pattern, namely the median ternary pattern (MTP) for face representation and recognition. The proposed MTP operator encodes the texture information of a local neighborhood by quantizing the intensity values of the neighbors around each pixel in three different levels based on the local median and a global threshold value. The distribution of the resulting MTP codes within an image or image patch is then used as the facial feature descriptor. The effectiveness of the proposed method has been evaluated with the FERET face image database using template matching. Experimental results show the superiority of the MTP feature descriptor in comparison to other well-known appearance-based methods.