2D-PCA Feature Extraction And Research About Its Application To Face Recognition

Face recognition technology has a wide range of application areas, such assearching for criminals in the public security department, dynamic monitoring andrecognising in the security department, monitoring password in bank system. Thestudy of face recognition and related technologies has important theoreticalsignificance and application value.This paper focuses on improving the rate of facial feature extraction andrecognition algorithm robustness then to build a set of human face recognition systembased on OpenCV. First, detecting the human face from the camera taken images orspecified image, then normalized and processed face images and use Eigenfacemethod for feature extraction, minimum distance classifiers to assort and then selectthe most closely matched the face from stored face database, and show the result. Inthis paper, the main tasks are as follows:Firstly, we have summarized the classic algorithm and advantage and weaknessof face detection and face recognition, based on the deeply analysis of face recognitiontechnology development at home and abroad. We are explicit of its difficult andsignificant points by grasping its application field and development trendSecondly, using the AdaBoost algorithm to detec face in order to determine itssize and location; Characteristics of different parts of the face have differentcontribution rates for face recognition; In this paper, we propose to improve thesub-block two-dimensional principal component analysis method, Furtherly improvethe robustness and effectiveness of the facial feature extraction and recognitionalgorithm; Then we use the minimum distance classifier for classification.Finally, this paper discusses the key issues to consider when designing facerecognition systems. Design and implement a face recognition system based onOpenCV.The system based on MFC-based platform, Visual Studio 2008 developmentenvironment, with OpenCV of the source function library. To build a small face database, we experiments on face recognition based on 10 people of the laboratory totest the performance of the system.