Research On Face Detection Algorithm Based On Adaboost And Neural Network

Face detection is a research subject stemming from the request of face recognition, with the development of computer technology face detection is researched independently. Not only is face detection the basis of face recognition and facial expression recognition, having influence on the research of these subjects, but also have applied widely in the area of public security and safe guard etc...As the first step of face recognition, the task of face detection is to detect human faces from background of image. However, face detection result is usually affected by the background, brightness or head posture of image and so on, which makes the process of detection more complicated.A great amount of literatures, surveys and research papers concerning up-to-date technique of face detection and face recognition are read and analyzed. Some hot issues about face detection are discussed and studied in this paper. The experiments indicate that the methods of face detection proposed in this paper are reasonable, showing a certain degree if theoretical and practical value. The research work of this paper mainly includes the following several aspects:1)About image processing, we make a discussion of a varity of basic image processing methods, that used to remove the interference informations in the face image, including gray normalization, scale normalization, median filtering and edge extraction etc...2)In the part of face detection, after the discussion of the traditional face detection methods, We mainly study the Face Detection method based on neural network and the principle of Adaboost, analysis the advantages and disadvantages of BP neural network in depth, at last, we use the BP neural network to improve the Adaboost, and get an outstanding Face Detection method-Face Detection methodbased on Adaboost and neural network.3)This paper finish a simple face detection system, and detailed design of the system for each step, and make a comparative analysis to the systems based on improved Adaboost and BP neural network classifiers and the cascade of traditional Adaboost.