The Study On Face Recognition Method Based On Geometry Graph

Face recognition was one of the most important subjects in pattern recognition inthe past twenty years. Numerous face recognition methods have been proposed. Currentsystems can do relatively accurate recognition under some restricted condition usingthese face recognition methods. It has attracted much attention to its potentialapplication values.Nowadays, face recognition has showed great application values in public safety,intelligent monitoring, digital identity authentication, E-commerce, digitalentertainment and so on. Requirements of the practicality of the face recognition arebeing increased as it is widely used. Also, most of the current face recognition methodscan do relatively accurate recognition in some fixed environments. However, there aresome issues in modern face recognition systems, such as long recognition time and vastcomputer memory occupation. Some academic achievements were got either:(1) Based on the history and content of the face recognition technology, theapplication and development of the face recognition technology were summarized. Themain face databases which are commonly used today and the one used in this paperwere introduced.(2) A face recognition algorithm was researched and proposed based on differentskin colors On the basis of the study of image preprocessing, image signal filterdenoising, gray-scale transformation, image binarization, edge detection, imagecropping, and face image normalized were realized.(3) In the process of feature extraction, construct a32feature vector, including the distance, angle, and area attribute.(4) In the face recognition part, the face property relationship diagram wasextracted to match purpose features and geometric information. The California instituteof technology in human face database on the experimental results shows that the methodis designed in this paper, using tree search algorithm advantages node correspondingrelation is clear, precision, but it is more time-consuming, storage capacity of the systemis also large. And greed and pruning algorithm is more economically in time, faster,storage capacity of the system is also relatively less, but it is less precision.