Improvement of Face Recognition Rate under Varying Illumination by Using Normalized Eigenface

It began to rely a lot on face detection and recognition techniques and use them in various applications in our daily lives, especially in recent years, which motivated companies and researchers to develop this technology and reduce its errors. Despite that, the results of these techniques are still affected by many factors, most notably the age, accuracy, and difference in illumination or the low lightness of the images. For this reason, we will create a system that reduces the effect of low and different illumination of the image and increases its recognition accuracy, by using the Eigenface algorithm (or PCA) with normalization to reduce errors brought on by background and lighting, also Histogram Equalization is used to enhance low-light images to achieving high recognition accuracy for low-light images and images that contains shadows. Finally, the experimental results on the Yale B, Extended Yale B and AR databases indicate that the proposed method is considering the best method to use in good and poor lighting cases, as it achieved the highest and most accurate results when used in all the databases. This system appears to be a successful way to solve some facial recognition problems and increases the rate and accuracy of recognition. The results show that the efficiency of the proposed work is enhanced by 13.586-222.967% compared with the Eigenface method and 0.389-10.294% compared with Normalize Eigenface method.