Fingerprint Indexing Based on Minutia-Centred Deep Convolutional Features

Most current fingerprint indexing systems are based on minutiae-only local structures which represent the relationships between the central minutia and its neighborhood. However, it is difficult to robustly extract minutiae from poor quality images, which significantly degrades the retrieval accuracy. To overcome this problem, this paper employs Deep Convolutional Neural Network (DCNN) to learn a minutia descriptor representing the local ridge structures. The learned Minutia-centred Deep Convolutional (MDC) features from one fingerprint are aggregated into a fixedlength feature vector by triangulation embedding method for the purpose of improving retrieval efficiency. In order to understand the MDC features, a steerable fingerprint generation method is proposed to verify that they describe the attributes of minutiae and ridges. Experimental results on two benchmark databases show that the proposed method achieves better performance on accuracy and efficiency than other prominent approaches.