Abstarct: The person recognition in video can be determined by a face recognition system. Using face recognition system at the entrance gate can be reduced violation and increased security. Face recognition in real-time scenarios is mainly affected by illumination variations, facial exxpression, pose variations, occlusion, scale variations, poor quality of the images and real-time system. Making the system robust to these problems is very difficult. Existing methods was focused on one of the above problems. Although the results is very good but they aren’t robust to other problems, so they are useless in real world. The purpose of this thesis is determine the identity of person in video entrance gate in real world. In this thesis, is present a real-time and robust method in different lighting, facial exxpressions, pose and scale variation, occlusion and poor quality of the images. In the first step video rcords from the camera, and the using local binary patterns and cascade classifier baxsed on Adaboost Asymmetrical for face extraction. In the next step, to choose the most appropriate faces, for of each person the extraction and separation of each individual face images. Select the most appropriate person is done to help four features, pose, illumination, bluring and resolution. After receiving a frxame, the choice of the individual images are sent to the phase identification. In the phase identification using SIFT descxriptor for feature extraction of prominent areas such as the eyes, nose and mouth are done and the size of these features is reduced. Finally identification is done with the help matching the features of input image with the features of images in the gallery. Experimental results on three databaxses Extended Yale B (illumination variation), FEI (pose variation) and ORL (facial exxpression and pose variation), indicating the superiority of the proposed method over other methods that attempt to identify individuals under different circumstances