One-Stage Mobile Palmprint Recognition via Keypoint Detection Network

With the increasing use of digital devices, researches on the security of user authentication have attracted considerable attention. Especially, on mobile devices, various information such as fingerprint and password are utilized for user authentication. However, these contact authentication methods cannot work well in certain environments, such as when the user's fingers are wet or dirty. In addition, by the COVID-19 pandemic, the need for contactless authentication methods has increased, and some studies introduced for contactless palmprint recognition on mobile devices. In this paper, we propose a one-stage palmprint recognition network with keypoint detection network for mobile named MPN. The proposed palmprint detection method is designed with keypoint detection network to predict the information needed for alignment, which is simpler and faster than the previous detection method. Then, we propose an adaptive ROI extraction method to align palm image and extract clear ROI data of the palmprint using the predicted information. By combining our proposed method, we develop a robust and fast network for palmprint recognition.