27 March 2019 in IEEE Transactions on Information Forensics and Security
Fingerprint authentication is widely used thanks to its simple process and low cost, but it is vulnerable to fake fingerprints. Many researchers have been working on presentation attack detection to ensure the security of fingerprint systems. However, the existing studies only focus on improving the detection accuracy, let processing time and memory requirement be out of focus. Hence, it is difficult to integrate the existing algorithms to embedded and mobile systems. This study proposes a method to detect presentation attacks using a small fully convolutional network. The proposed network is designed using the structure of Fire module of SqueezNet. The use of Fire module results in a network which has around 0.5 million parameters. The proposed network is trained using images of 32×32,48×48, or64×64 pixels. Since the network has no fully connected layer, it can do inference with images of any size. This advantage helps to improve the detection rate and allows the proposed algorithm to be integrated into fingerprint systems easily. The experiments show an average detection error of 1.43%, which is comparable with the state-of-the-art accuracy, while the processing time and memory requirement is much reduced.
13 December 2019 in IET Biometrics
Minutiae, widely used feature points of fingerprint images, directly decide the performance of fingerprint recognition. Conventional minutiae extractors rely on a series of preprocessing steps, thus performing poorly for bad quality samples due to error accumulations. Existing extractors using convolutional neural networks are trained and tested with a certain specific sensor, thus requiring various modules for different sensors. To solve these problems, a universal minutiae extractor using a modified U‐shaped segmentation network is proposed. Specifically, the proposed extractor classifies each pixel of a fingerprint image into a category of minutia with a certain orientation or non‐minutia point, thus obtaining location and orientation information of minutiae simultaneously. The experimental results plus comparisons with other academic and commercial extractors prove that the proposed network can extract accurate and robust minutiae regardless of the quality of fingerprints and the sensor types.