Robust Clustering-based Segmentation Methods for Fingerprint Recognition

Abstract

Fingerprint recognition has been widely studied for more than 45 years and yet it remains an intriguing pattern recognition problem. This paper focuses on the foreground mask estimation which is crucial for the accuracy of a fingerprint recognition system. The method consists of a robust cluster-based fingerprint segmentation framework incorporating an additional step to deal with pixels that were rejected as foreground in a decision considered not reliable enough. These rejected pixels are then further analysed for a more accurate classification. The procedure falls in the paradigm of classification with reject option-a viable option in several real world applications of machine learning and pattern recognition, where the cost of misclassifying observations is high. The present work expands a previous method based on the fuzzy C-means clustering with two variations regarding: i) the filters used; and ii) the clustering method for pixel classification as foreground/background. Experimental results demonstrate improved results on FVC datasets comparing with state-of-the-art methods even including methodologies based on deep learning architectures.

Dual Spectral Attention Model for Iris Presentation Attack Detection

“…The work of [21] made the model less reliant on presentation attack samples by requiring it to learn from real samples. Despite the small size of the dataset, the authors found that their model achieved better loss and accuracy, as well as lower error rates, in detecting attacks and real iris samples, despite the dataset's limited size.…”

A Novel Dynamic Fingerprint Segmentation Method Based on Fuzzy C-Means and Genetic Algorithm

“…But the segmented foreground area was slightly different from the benchmark according to the segmentation results in their paper. And in [5], an improved method for fingerprint segmentation based on the method, presented in [6], was applied. In order to make an improvement, they focused on filters and pixel clustering classification.…”

“…There still remain several limitations in the aforementioned approaches [19]. First, most existing clustering-based algorithms need to be trained in advance (e.g., [3], [7] and [15]) or require prior knowledge about the number of clusters, showed in [5], [6] and [12]. Since it's hard for online AFIS to obtain the number of clusters, those methods have difficulty in deploying their framework in practice.…”