Study On Face Recognition Under Unlimited Conditions

The traditional authentication methods, such as" user name+password", smartcards, certificates and so on, have not met the actual safety requirements with the rapiddevelopment of modern information technology. Especially after the United Statessuffered911terrorist attack, the technology of face-recognition by the internationalacademic circles and business circles paid close attention to. Human wealth ofphysiological characteristics (such as fingerprint, face, iris) and behavioralcharacteristics (e.g., handwriting, voice, gait) as human inherent can not copy thephysiological characteristics, for authentication to provide a safer, securer and moreconvenient solution. Among them, the technology face-recognition has the advantagesof non-contact, non-mandatory, concurrency and others relatives to other biologicalrecognition technology. In the field of pattern recognition and image processing, it is avery active research topic. Under the limited conditions, the recognition rate can reachmore than95%, but under the unlimited conditions, the recognition rate will decreaserapidly and can not meet the actual applicative demands.The main work and research results of this paper are as follows:1．Use ODVS(Omni-Directional vision sensor) and high speed dome camera totarget the human body get monitoring scene object detection and tracking, eventuallygain can be used for face recognition images. ODVS can get all-round visual imagerange, high speed dome camera through camera and ball machine suitably, can controlcamera to need free rotation Angle, access to human targets under the differentperspectives of multiple head image. And these images can be used to identify you offace image, or how many tickets can be used for identification of face image,need touse face detection algorithm is determined. If no can be used for identification of faceimage, it also fast control high speed dome cameras are photographed film, again untilthe face image can be used to identify.2. This paper introduces the weighted block in two directions of the twodimensional principal component analysis (Modular Weighted(2D)~2PCA, MW- (2D)~2PCA)for face recognition.This method can effectively extract the face of localfeature, based on the image processing block reduces the posture, light and otherexternal environment on face recognition result influence,Besides，the fuzzy theory isintroduced to classify the single sample face recognition. The experimental resultsshow that this method has a high recognition rate and good robustness.