Research on ear recognition is a new subject in the field of biometrics authentication. This paper presents a recognition approach for ear biometric based on fusing local binary pattern (LBP) and Gabor filter. The local Gabor binary pattern (LGBP) descriptor is utilized to extract the texture feature of ear image and K-NN classification is adopted. The proposed approach is tested on the USTB ear image set. The experiment results show that the scheme is effective and promising. The recognition correct rates based on LGBP are better than those based on LBP and the ear recognition approach based on LGBP using Chi square measure yields an exciting recognition rate of 98.71%.