青教师论坛.pptxVIP

An Introduction of Data Fusion Technology with Applications in Biometric Person Recognition;个人简历;Data Fusion Applications (1);Data Fusion Applications (2);Data Fusion Applications (3);Data Fusion as a General Problem;Data Fusion as a General Problem;Data Fusion as a General Problem;Data Fusion as a General Problem;Data Fusion as a General Problem;Data Fusion as a General Problem;Data Fusion as a General Problem;An Example of Data Fusion: Combination of Person Recognition Classifiers;An Example of Data Fusion: Combination of Person Recognition Classifiers;Combination of Classifiers: Basic Definitions;Combination of Classifiers: Basic Definitions;Combination of Classifiers: The Problem;Fusion of Classifiers: Fusion Function;An Example of Data Fusion: Combination of Person Recognition Classifiers;An Example of Data Fusion: Combination of Person Recognition Classifiers;An Example of Data Fusion: Combination of Person Recognition Classifiers;We use Mel cepstral coefficients (MFCC) derived from a mel-frequency filterbank to represent the features for identification. Speaker modeling is based on the Gaussian mixture model (D. A. Reynolds and R. C. Rose 1995). A Gaussian mixture density is a weighted sum of M component densities: Each component density is a D-variant Gaussian function of the form: with mean vector and covariance matrix . The speaker identification method is tested by the first CD of set CDS003 of XM2VTS (K. Messer et al. 1999) database, which contains speech files of 74 people and each speaker provides 24 speech files. We use the first 12 files of 24 speech files as the training set and the other 12 files as the testing set. A correct identification rate of 87.0% is obtained. ;An Example of Data Fusion: Combination of Person Recognition Classifiers;An Example of Data Fusion: Combination of Person Recognition Classifiers;Face Detection: Haar-like Features (Viola and Jones, 2004).;Efficient Algorithm for Feature Calculation;The most representative H