数字图像处理第十六章.ppt

* * * * * * * * O O Face recognition can be categorized into appearance-based, geometry-based, and hybrid approaches. * K-means The main advantages of this algorithm are its simplicity and speed which allows it to run on large datasets. Its disadvantage is that it does not yield the same result with each run, since the resulting clusters depend on the initial random assignments. It minimizes intra-cluster variance, but does not ensure that the result has a global minimum of variance. Discussion An example: Using k-means methods clustering the following 9 points into 3 classes: (-1,1)，(0,1)，(1,1)，(1,0)，(2,0)，(-1,0)，(-1,-1)，(0,-1)，(0,-2). Assuming the first three points to be the initial center of 3 classes. K-means An Application: Biometrics 3.14159 Security Level What you have What you know What you are/ What you do Method What you have? Possession (Key, Credit card, …) What you know? Knowledge (password, PIN, …) What you are/do? Characteristics (fingerprint,iris, …) Universality: Every person should possess this characteristic Uniqueness: No two persons possess the same characteristic Permanence: Does not change in time. i.e., it is time invariant; Long vs. short-term stability. Collectability: Can be quantitatively measured Non-intrusive Reliable and robust Cost effective for a given application Biometrics ? Access control Protection of security information ? Law enforcement The identification of know criminals ? Welfare agency Track control benefit distribution ? Bank credit card for check cashing ? Health care Track access to medical information Applications of Biometrics Taxonomy of Biometrics Technology Others (Ear, Retina, Dental) Biometrics Physical Behavioral Head Hand Signature Voice Others (Gesture, gait, Keystroke) Face Iris Fingerprint Hand Geometry Palmprint Biometrics ??????????????????????????????????????????????????????????????? Biometrics ? The same person may have the different signatures Challenges of