| dc.description.abstract | Pattern recognition is a scientific discipline for the classification of objects into a number of categories or classes. Today, one of the intensive researches in the field of pattern recognition is face recognition. In this research, we tried to perform face recognition by using Voting Feature Interval 5 (VFI5) classification algorithm. VFI5 algorithm is an algorithm that represents a description of a concept by a set of feature intervals and performs classification based on the voting feature. Features used in this research were generated from gray level face images. The face image dimension is 112×92 pixels which means there will be 10304 features which caused difficulties to achieve better accuracy with less storage and computational complexity. For that, Principal Components Analysis is used to reduce the dimensions. This is done by projecting an image on eigenface. Eigenface is obtained by projecting training data on eigen vector. Experiments were performed by using cross validation. Ten fold cross validation was used as cross validation technique with 10 repetition of single observation from the original sample as the validation data and the remaining as training data. The result of 10 observations subset are respectively 80%, 100%, 90%, 100%, 90%, 100%, 100%, 90%, 100% and 90%. Resulting in an average accuracy of 94%. We conclude the VFI5 algorithm good at recognizing face images. | id |
