A New Multi Fractal Dimension Method for Face Recognition with Fewer Features under Expression Variations

In this work, a new method is presented as a mingle of Principal Component Analysis (PCA) and Multi-Fractal Dimension analysis (MFD) for feature extraction. Proposed method makes use of best decision taken from both the methods and make use of fewer and effective features than traditional algorithms without compromise in recognition accuracy. In order to ease the pre-processing we controlled the variance in each mode. It is done to train the system to understand and recognize facial variance present in image. Experiments with different datasets show that proposed method is more suitable for larger dataset, with higher efficiency.