Research On The Method Of Kinect AAM Feature Point Location

Facial feature point localization is an advancing research in Computer Vision study which has many kinds of research areas,whose research results are used for human face detection,posture acquisition,expression micro-detection,human-computer interaction,cognitive psychology and so on.And it also provides technological support to other business areas,such as network identity security,online payment security,movie special effects scene simulation,the new human-computer interaction interface and so forth.However,due to the complexity of the human facial structure,which means it is not easy to accurately identify,it has still a long way to go to perfectly solve the problem of facial features.The key points of facial features are based on the eye,mouth,eyebrows,as well as the image analysis and extraction of outline profile.First of all,in this paper,in using the Kinect facial feature point extraction,the key parts of human facial extraction are in accuracy and robustness and the main matching error comes from the results of illumination instability and attitude transformation.Because the instability of the illumination has less interference to the three-dimensional image,meanwhile,the accuracy of the language attitude judgment with the depth system is also enhanced;hence,this device is used to acquire the image in this paper.In this paper,it is proposed to analyze the theoretical derivation of Eigenface and Fisherface and their performance feasibility in practice.The two algorithms are applied to the ORL and Yale databases respectively for performance simulation experiments.Finally,a new face recognition system is proposed,which tries to analyze and implement each part of the theory in this method.And an improved AAM facial feature localization model is established.The Gabor transform and Fisherface algorithm are used to extract and recognize the facial features.At last,an improved AAM model fitted by the human face to do the facialfeature accurately.Through the use of MFC programming framework,a face recognition system is established in this paper with a set of strong anti-interference abilities,including the amount of light changes,the attitude of the random simulation extraction,micro-expression capture,light intensity changes undisturbed,strong robust.