A multi-model restoration algorithm for recovering blood vessels in skin images

Abstract

Blood vessels under skin surface have been used as a biometric trait for many years. Traditionally, they are used only in commercial and governmental applications because infrared images are required to capture high quality blood vessels. Recent research results demonstrate that blood vessels can be extracted directly from color images potentially for forensic applications. However, color images taken by consumer cameras are likely compressed by the JPEG compression method. As a result, the quality of the color images is seriously degraded, which makes the blood vessels difficult to be visualized. In this paper, a multi-model restoration algorithm (MMRA) is presented to remove blocking artifacts in JPEG compressed images and restore the lost information. Two mathematical properties in the JPEG compression process are identified and used to design MMRA. MMRA is based on a tailor-made clustering scheme to group training data and learns a model, which predicts original discrete cosine transform coefficients, from each grouped dataset. An open skin image database containing 978 forearm images and 916 thigh images with weak blood vessel information and a set of diverse skin images collected from the Internet are used to evaluate MMRA. Different resolutions and different compression factors are examined. The experimental results show clearly that MMRA restores blood vessels more effectively than the state-of-the-art deblocking methods.