Based On Two Adaboost Lbp Fast Face Recognition Implementation And Application,

Face recognition which used some methods to extract characters from the face image is the key for the automatic computer vision; it is also the hot issue in the pattern recognition. There are many fields we can use the face recognition, including content searching, identity recognition. Now a set of algorithms are used in this domain, but lack of application in reality. Some methods are very complex which is hard to compute, some methods just apply to fix data. The precondition of face recognition is the detection of face region out of the face image. So the key factor is the time, if we can recognize the face from the image in real time, it can bring bright future for the application. In this paper we bring forward a new method which use two levels Adaboost and LBP, in the experiments, we prove that it can achieve high correct rate more than 90% and recognize the face in real time.The paper has two phases, the first phase is the face detection and key regions detection, we get the key regions from the result which get from the face detection. The second phase is the expression of these key regions.In the first phase we use the Adaboost to build the cascade classifier of face, use the cascade method we can filter the input image regions, a lot of regions can be swept out. And also use Adaboost to build the classifier of key regions of face. In the second phase, use the LBP method to get the characters from these regions. Last we use the chi-square measure to find the most matching sample in database. In this paper we bring the weight into the comparison to distinguish the contributions of different regions. The most important character of this method it that the database can extend at any time, insert an sample into the database will not result in recomputing of all samples in the database.