Image quality treatment to improve iris biometric systems

Abstract

Currently, there are many solutions related to biometric systems for users authentication purposes based on iris image, because, the human iris characteristics remain unalterable for each individual. In order to improve the iris biometric system accuracy, the images captured need to reach a minimum quality level. However, different distortion can occur in the image acquisition step, such as blurring, affecting the global performance of the system. In this context, this research proposes a methodology to treat the blurred images and to improve their quality. The deblurring process is a complicated task, and most of the methodologies need the Point Spread Function (PSF) to correct the impaired image. A method named blind deconvolution is used in this work to get the PSF value, and then it is used in the deblurring process to improve the image quality. Experimental results demonstrated that corrected images reached minor Hamming Distance values, considering iris images of the same user, improving the biometric system performance. Thus, with the solution proposed, the False no-Match rate parameter of the global system decreased from 27.82% to 4.41%.