Image Quality Assessment for Effective Ear Recognition

Due to the recent challenges in access control, surveillance and security, there is an increased need for efficient human authentication solutions. Ear recognition is an appealing choice where the data acquisition procedure is contactless, non-intrusive, and covert. This article proposes a deep learning-based solution for effective ear recognition. First, we explore multiple strategies to enhance learning using alternative ear datasets with a wide range of ear poses. Second, we investigate the performance of the proposed deep ear models in the presence of various image artifacts, which commonly occur in real-life recognition applications, to identify the robustness of the proposed ear recognition models in controlled and uncontrolled conditions (dataset dependent). Finally, we propose an efficient ear image quality assessment tool designed to guide the proposed ear recognition system. By performing a set of experiments on extended degraded ear datasets, we determine that the employment of the proposed ear image quality assessment tool improves ear identification performance from 58.72% to 97.25% for the USTB degraded dataset and from 45.80% to 75.11% for the degraded FERET dataset.