Noise suppression methods for low quality images with application to face recognition

Any automatic face recognition system (FRS) operating under uncontrolled environment with no user interaction may encounter difficulties in its performance when low quality image is received as input. This issue may require the necessity of employing some preprocessing steps to enhance the image quality. One example of low quality image is provided by noisy data. The impact of six approaches for noise suppression with respect to the face recognition task is addressed in this paper. Two noise types namely, Gaussian and “Salt & Paper”, respectively are added to the noise-free images to degrade the image quality. Surprisingly, although important, to the best of our knowledge, no systematic comparison exists in the literature with respect to this particular issue. Apart from the recognition enhancement capability, each noise suppression algorithm is analyzed with respect to the processing time, which is also crucial for a complex FRS operating in real time.