Enhanced Tattoo Image Quality Assessment Through Multispectral Sensing

The impact of using tattoo images in biometrics and forensics-related applications for human identification or in the case of ruling out suspects has been significantly increased in recent years. The NIST Tatt-C and Tatt-E biometric tattoo challenges have focused on pushing forward tattoo related research in terms of human identification, detection, localization, and sketch-matching. However, current tattoo research has been limited to the use of the visible wavelengths (380-750 nm) when imaging tattoo skin regions. Variations in skin pigments, ink colors, and ink types make the imaging of tattoos inherently challenging when operating in a conventional band only. For example, enhanced imaging techniques are needed to improve the contrast between skin and tattoo regions in individuals with darker skin pigment or in instances where the ink color is close to the skin pigment. In this article, we propose the use of various wavelengths within the shortwave infrared (SWIR) spectrum that provide enhanced tattoo imaging and, as a result, an enhanced human identification capability. First, we performed a data collection process that resulted in the generation of a large scale database of visible and SWIR tattoo images for 133 subjects (Session 1) with 80 subjects coming back for a second round of data collection (Session 2). Furthermore, we conducted an empirical study using various skin pigments (determined via known ethnic group meta data) and when image quality varies within and across different SWIR wavelengths. Experimental results indicate that specific SWIR wavelengths, namely the range between 1100 nm and 1300 nm, show superior tattoo image quality in the majority of the ethnic groups studied, except one (i.e., Asian group). To the best of our knowledge, this study is the first attempt to bring forward the benefits of tattoo imaging outside the visible band for the purpose of human identification.