Face Recognition System And Face Detection Algorithm

Information has become the contemporary society one of the most important strategic resource. Currently, as a high level of security, including face recognition, will be disruptive in more areas.The most popular face recognition focused on two-dimensional image . Some algorithms are mostly targeted at the exploration of local features.The difference specifically the face with the different temperature emit the infrared spectroscopy; In complicated environment, accurate detecting goes out of person face from different lighting, distance, angle.This design thought in reference to much of the literature concluded after a new face monitoring system based on a large database. Can be considered from a massive database, using clustering method to classified the value, calculated the properties of respective classes. Finally,Complete image which to be identified with the subset.Main components of the database including the international database. You can also use mirror transformation and micro-degree rotation to meet the vast amounts of data. Preprocessing including denoising, recovering the fuzzy motion picture, histogram equalization, two-dimensional wavelet transform to compensate for illumination, median filtering. Feature extraction by the ways of PCA and ICA, the values are stored after scanning. Thus end of the formation of the entire database. To determine whether the image to be detected, introduce the idea of face detection.This paper presents a method which is based on template matching using the Canny operator to extract edges and then traverse to cut out the face. Finally, in accordance with the similarity of all the cluster center, after showing the most similar class, you can arrive at the classification results.K means algorithm, simulated annealing K-means algorithm, and genetic algorithm method is used to achieve the purpose.This exploration and realization of the entire face recognition system, from the perspective of theory and practice, and reduce the computational time, improve the retrieval efficiency.