Abstract
In this paper, a novel face description method called a completed model of local ternary pattern (CLTP) for face representation and recognition. Firstly, the original face image is divided into small blocks from which the local difference value sign-magnitude transform was computed. Extracting the histogram statistical characteristics of each block by CLTP_P, CLTP_N and CLTP_M, and information entropy of the sub blocks are calculated separately. Then the histograms of all the blocks are linked to get the CLTP feature to be used as the final face descriptor. Finally, .the k- nearest neighbor classifier with chi-square is used for classification and recognition. Experimental results on AR and Yale face databases show that the proposed method has good face representation under different facial expression, partial occlusion, different illumination. Also the recognition rate is higher than LTP and variations of LTP method.
Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
