A Code-Level Approach to Heterogeneous Iris Recognition

摘要: Matching heterogeneous iris images in less constrained applications of biometrics is becoming a challenging task. The existing solutions try to reduce the difference between pixel intensities or filtered features. In contrast, this paper proposes code-level approach recognition. non-linear relationship binary feature codes modeled by an adapted Markov network. This model transforms number templates probe into homogenous template corresponding gallery sample. addition, weight map on reliability can be derived from model. learnt and are jointly used building robust matcher against variations imaging sensors, capturing distance, subject conditions. Extensive experimental results matching cross-sensor, high-resolution versus low-resolution and, clear blurred demonstrate achieve highest accuracy compared with pixel-level, feature-level, score-level solutions.