A Reinforcement Learning Agent for Minutiae Extraction from Fingerprints

In this paper we show that reinforcement learning can be used for minutiae detection in fingerprint matching. Minutiae are characteristic features of fingerprints that determine their uniqueness. Classical approaches use a series of image processing steps for this task, but lack robustness because they are highly sensitive to noise and image quality. We propose a more robust approach, in which an autonomous agent walks around in the fingerprint and learns how to follow ridges in the fingerprint and how to recognize minutiae. The agent is situated in the environment, the fingerprint, and uses reinforcement learning to obtain an optimal policy. Multi-layer perceptrons are used for overcoming the difficulties of the large state space. By choosing the right reward structure and learning environment, the agent is able to learn the task. One of the main difficulties is that the goal states are not easily specified, for they are part of the learning task as well. That is, the recognition of minutiae has to be learned in addition to learning how to walk over the ridges in the fingerprint. Results of successful first experiments are presented.