Research Of Face Recognition Based On SIFT Algorithm

Face recognition is a kind of computer technologies, which described the biological characteristics of human face and distinguish a face from others by computer. A person’s mood, race, identity information can be inferred by the face. Facial recognition technology usually refers to judge people’s identity after the human face feature extraction, feature analysis and similarity calculation. To be a computer security technologies, facial recognition technology used widely to the public security, national defense and other important field of information security.Scale invariant feature transform is put forward in recent years and the hot issues in the field of digital image signal processing. Has been widely used in object detection, motion trajectory, the human body tracking, image matching, image recognition and other fields.in this paper mainly studied the SIFT algorithm, and apply it to the face recognition, some improvements were made on the basis of the original algorithm, and proposes a face recognition method based on SIFT.This paper introduce the SIFT algorithm, expounds the structure of the scale space, SIFT feature point detection, the description of the feature points and SIFT feature matching. The algorithm is complicated, not easy to understand and has overmuch operation, while it has the huge computational cost which brings great difficulty to the practical application. Improved the description of SIFT feature points and matching algorithm, this paper proposes a face recognition method based on SIFT. Reduces the dimension of SIFT descriptor after improving the SIFT feature point description, thus reduce the matching time. By our matching method, reduce the search scope, improve the matching accuracy.This algorithm is also simulated by VS2008, through the subjective visual effect and objective experimental data, the presented algorithm is analyzed and evaluated. The experimental results verify the efficiency of the proposed algorithm.