Computerized system for fingerprint classification using singular points

As one of the most important and obvious global features for fingerprints, the singular point plays an essential role in fingerprint registration and fingerprint classification. To date, the singular point detection methods in the literature can be generally divided into two categories: methods based on traditional digital image processing and those on deep learning. Generally speaking, the former requires a high-precision fingerprint orientation field for singular point detection, while the latter just needs the original fingerprint image without preprocessing. Unfortunately, detection rates of these existing methods, either of the two categories above, are still unsatisfactory, especially for the low-quality fingerprint. Therefore, regarding singular point detection as a semantic segmentation of the small singular point area completely and directly, we propose a new customized convolutional neural network called SinNet for segmenting the accurate singular point area, followed by a simple and fast post-processing to locate the singular points quickly. The performance evaluation conducted on the publicly Singular Points Detection Competition 2010 (SPD2010) dataset confirms that the proposed method works best from the perspective of overall indexes. Especially, compared with the state-of-art algorithms, our proposal achieves an increase of 10% in the percentage of correctly detected fingerprints and more than 16% in the core detection rate.

A robust singular point detection algorithm

Fingerprint has been adjudged as the most reliable means of identification and authentication of an individual because of its uniqueness, high immutability and unchanging patterns. Typically, the ridges of fingers consist of different pattern types whose attributes are often extracted for matching during fingerprint classification. The occurrence of these patterns vary among fingerprint, which has significantly affected the accuracy of fingerprint recognition and classification processes. This paper therefore focuses on the investigation of the impact of pattern variation, singular points and feature points on fingerprint matching. The investigation was based on benchmarked FVC2000, FVC2002, FVC2004 and FVC2006 fingerprint databases which comprise four datasets each from different sources and of varied types. The obtained false non match rate (FNMR), false match rate (FMR), total matching time (TMT) and average matching time (AMT) values revealed that use of multiple matching criteria will lead to extension in the fingerprint matching time.

It is a very demanding task to design a reliable fingerprint matching approach with high accuracy and speed. An algorithm based on the novel structure combining the singular point with its neighborhood minutiae proposed in this paper can solve this problem efficiently. The structure introduced in this paper has two novel ideas as follows: First, we give an efficient singular points detection method by the inter-relationship between the singular points and the minutiae around them. It can reject the spurious singular points detected by the Poincare index. Second, an improvement of minutiae pairing strategy is introduced, which can sharply decrease the number of candidate minutiae pairs. Experiment results show that these improvements can highly speed up the matching with a preferable accuracy. This algorithm can be used in the one-to-many matching of the on-line fingerprint identification system.KeywordsFingerprintMatchingSingular PointMinutiae Pairing

A variational formulation for fingerprint orientation modeling

Automatic matching of poor quality latent fingerprints to rolled/slap fingerprints using an Automated Fingerprint Identification System (AFIS) is still far from satisfactory. Therefore, it is a common practice to have a latent examiner mark features on a latent for improving the hit rate of the AFIS. We propose a synergistic crowd powered latent identification framework where multiple latent examiners and the AFIS work in conjunction with each other to boost the identification accuracy of the AFIS. Given a latent, the candidate list output by the AFIS is used to determine the likelihood that a hit at rank-1 was found. A latent for which this likelihood is low is crowdsourced to a pool of latent examiners for feature markup. The manual markups are then input to the AFIS to increase the likelihood of making a hit in the reference database. Experimental results show that the fusion of an AFIS with examiner markups improves the rank-1 identification accuracy of the AFIS by 7.75% (using six markups) on the 500 ppi NIST SD27, 11.37% (using two markups) on the 1000 ppi ELFT-EFS public challenge database, and by 2.5% (using a single markup) on the 1000 ppi RS&A database against 250,000 rolled prints in the reference database.

Fingerprint analysis is typically based on the location and pattern of detected singular points in the images. These singular points (cores and deltas) not only represent the characteristics of local ridge patterns but also determine the topological structure (i.e., fingerprint type) and largely influence the orientation field. A core-delta relation is used as a global constraint for the final selection of singular points. This paper presents and discusses rules for implementing a structural fingerprint classifier. The proposed approach is divided in two main phases: Singular points extraction and recognition algorithm. The fingerprint classification is heavily depended on the singularities, i.e., core point and delta points. For the recognition of fingerprint a new technique is proposed based on the number of singular points and their types.

Fingerprint classification is used to assign fingerprints into five established classes, namely, Whorl, Left loop, Right loop, Arch, and Tented Arch, on the basis of ridge structures and singular points’ trait. Although some progresses have been achieved in improving accuracy rates, problems arise from ambiguous fingerprints, especially those with large intraclass and small interclass variations. Poor-quality images, such as those with blur, dry, wet, low contrast, cut, scarred and smudgy features, are equally challenging. This study proposes a new classification technique based on template matching by using fingerprint salient features as a matching tool. In classification phase, a new set of fingerprint features is created based on singular points’ occurrence and location along the symmetric axis. A set of five templates, in which each template represents a specific true class, is then generated. Finally, classification is performed by calculating similarity between the query fingerprint image and the template images by using ×2 distance measure. The performance of the current method is evaluated in terms of accuracy by using 27,000 fingerprint images acquired from the National Institute of Standard and Technology Special Database 14, which is a de facto dataset for development and testing of fingerprint classification systems. Experimental results have a high accuracy rate of 93.05%.

An algorithm is proposed which combines Zero-pole Model and Hough Transform(HT) to detect singular points. Orientation of singular points is defined on basis of Zero-pole Model which can further explain the practicability of Zero-pole Model. Contrary to orientation field generation, detection of singular points is simplified to determine the parameters of Zero-pole Model. HT uses rather global information of fingerprint images to detect singular points. This makes our algorithm more robust to noise than methods which only use local information. As Zero-pole Model may have a little warp from actual fingerprint orientation field, Poincare index is used to make position adjustment in neighborhood of the detected candidate singular points. Experimental results show that our algorithm performs well and fast enough for real time application in database NIST-4.

Analysis of traffic flow signals plays an important role in traffic prediction and management. As an intrinsic property, the singular point of a traffic flow signal labels a new nonsteady status. Therefore, detecting the singular point is an effective approach to determine the moment of traffic flow prediction. In this paper, an improved wavelet transform is proposed to detect singular points of real-time traffic flow signals. The number of detected singular points is output via the heuristic selection of multiple scales. Then, a weighted similarity measurement of historical traffic flow signals is utilized to predict the next singular point. The position of the next singular point decides the duration of prediction adaptively. The detected and predicted singular points are applied to dynamically update the unequal interval prediction of traffic flow. Furthermore, a Vasicek model is used to predict the traffic flow by minimizing the sum of the relative mean standard error (RMSE) between the traffic flow increment in the predicted interval and the sampled increments of previous intervals. A decomposition method is used to solve the unequal matrix problem. Based on the scenario and traffic flow imported from the real-world map, the simulation results show that the proposed algorithm outperforms existing approaches with high prediction accuracy and much lower computing cost.

An important step in automatic fingerprint identification system (AFIS) is fingerprint matching. The task of fingerprint matching is to verify whether two fingerprints are coming from same finger. In this paper we detail and discuss the fingerprint matching algorithm. A minutia matching algorithm is proposed which modified the algorithm presented by Jain et al. In this algorithm, in order to reduce the effect of noise and false minutiae, block orientation and ridge information are introduced into the minutiae-based matching algorithm in a simple but reliable way. Ridge information which is some sampled points in the ridge are used to align two fingerprints, in order to avoid misalign two fingerprints, we use block orientation to correct the ridge alignment. Experiments on the database FVC2002 show the performance.

A majority of the minutiae based fingerprint verification algorithms rely on explicit or implicit alignment of the minutiae points for matching the two prints. With no prior knowledge about point correspondences, this becomes a combinatorial problem. Global features of the fingerprints such as the core and delta points represent intrinsic points of reference that can be used to align the two prints and reduce the computational complexity of the matcher. However, automatic extraction of singular points is usually error prone and is therefore not used by existing matchers. But, a systematic study of the impact on matching performance when core/delta points are available has not been done to date. In this paper, we explore the effects of the availability of reliable core and delta points on speed and accuracy of a matching algorithm. Towards this end, we present significant improvements to core and delta point detection algorithm based on complex filtering principles originally proposed by Nilsson et al., (2005). We also present a modified graph based matching algorithm that can run in O(n) time when the reference points are available. We analyse the resulting improvement in computational complexity and present experimental evaluation over FVC2002 database. We show that there is upto 43% improvement (70.2 ms to 39.8 ms) in average verification time and almost no loss in accuracy when reliable core and delta points are used.

Removal of digitization errors in fingerprint ridgelines using B-splines

This paper focuses on a general framework for singular point extraction from vector field. We design a new index of singular point based on complex polynomial model. We test our method in the publicly available benchmark dataset of the singular point detection competition (SPD2010). Our algorithm gets the best results and produces large margins compared to the top five algorithms which took part in the public competition. We also compare our algorithm with the state-of-the-art singular point detection algorithm (called "ZPM" method) with the benchmark. The performance of our method is much better than that of the state-of-the-art method.

Fingerprint classification helps in reducing the number of comparisons during the matching stage in automatic fingerprint identification system. In this study, a convolutional neural network model is proposed for classification of plain, rolled and latent fingerprints. We first propose a new convolutional neural network model initialised with random weights and train the model on fingerprint images. Then we fine-tune two pre-trained convolutional neural network models on fingerprint images. Finally, we compare these three models: two pre-trained models and a custom convolutional neural network model initialised with random weights. We show that pre-trained models can achieve the state-of-the-art results on other similar tasks with no or little fine-tuning. We also show that training data size and depth of the network have a serious impact on the overall performance of deep networks. Dropout is used to enhance the performance of deep networks where the labelled training data is not of sufficient size. All the three models trained on NIST DB4 fingerprint and IIIT-D latent fingerprint databases report good accuracy. By only fine-tuning the pre-trained convolutional neural network model, we get the accuracy of 99%, easily out-performing the state-of-the-art.