Image Band-Distributive PCA Based Face Recognition Technique

This paper presents an Image Band-Distributive PCA (IBD-PCA) based technique for face recognition. The proposed method consists of four steps. In the first step, the reference image is pre-processed by converting its pixel values and performing histogram equalization to increase its contrast. In the second step, the equal-size boundary calculation method is used to calculate the boundary splitting values to divide the input image into multiple images with respect to band intensities of pixels. In the third step, Principal Component Analysis (PCA) is used to extract features from the images which will then be used as the input for the fourth step. In the last step, matching is performed by calculating the Euclidean distance between principal components. The proposed technique has been tested on the ORL Face Database and Yale Face Database. The experimental results demonstrate that the proposed technique outperforms other techniques on the same database.