En Zhu

A systematic method for fingerprint ridge orientation estimation and image segmentation

This paper proposes a scheme for systematically estimating fingerprint ridge orientation and segmenting fingerprint image by means of evaluating the correctness of the ridge orientation based on neural network. The neural network is used to learn the correctness of the estimated orientation by gradient-based method. The trained network is able to distinguish correct and incorrect ridge orientations, and as a consequence, the falsely estimated ridge orientation of a local image block can be corrected using the around blocks of which orientations are correctly estimated. A coarse segmentation can also be done based on the trained neural network by taking the blocks of correctly estimated orientation as foreground and the blocks of incorrectly estimated orientation as background. Besides, following the steps of estimating ridge orientation correctness, a secondary segmentation method is proposed to segment the remaining ridges which are the afterimage of the previously scanned fingers. The proposed scheme serves for minutiae detection and is compared with VeriFinger 4.2 published by Neurotechnologija Ltd. in 2004, and the comparison shows that the proposed scheme leads to an improved accuracy of minutiae detection.

Two steps for fingerprint segmentation

A fingerprint image usually consists of different regions: non-ridge regions, high quality ridge regions, and low quality ridge regions. Fingerprint segmentation is usually to exclude non-ridge regions and unrecoverable low quality ridge regions as background so as to avoid detecting false features. In ridge regions, including high quality and low quality, there are often some remaining ridges which are the afterimage of the previously scanned finger and are expected to be excluded as background. However, existing segmentation methods do not take this case into consideration, and often, the remaining ridge regions are falsely taken as foreground. This paper proposes two steps for fingerprint segmentation to exclude the remaining ridge region from the foreground. The non-ridge regions and unrecoverable low quality ridge regions are removed as background in the first step, and then the foreground produced by the first step is further analyzed so as to remove the remaining ridge region. The experimental results showed the effectiveness of the proposed method in segmenting the remaining ridges as background and in turn producing much less spurious minutiae than the existing method.

An Incremental Feature Learning Algorithm Based on Least Square Support Vector Machine

Incremental learning has been widely addressed in machine learning literature to deal with tasks where the learning environment is steadily changing or training samples become available one after another over time. Support Vector Machine has been successfully used in pattern recognition and function estimation. In order to tackle with incremental learning problems with new features, an incremental feature learning algorithm based on Least Square Support Vector Machine is proposed in this paper. In this algorithm, features of newly joined samples contain two parts: already existing features and new features. Using historic structural parameters which are trained from the already existing features, the algorithm only trains the new features with Least Square Support Vector Machine. Experiments show that this algorithm has two outstanding properties. First, different kernel functions can be used for the already existing features and the new features according to the distribution of samples. Consequently, this algorithm is more suitable to deal with classification tasks which can not be well solved by using a single kernel function. Second, the training time and the memory space can be reduced because the algorithm fully uses the structural parameters of classifiers trained formerly and only trains the new features with Least Square Support Vector Machine. Some UCI datasets are used to demonstrate the less training time and comparable or better performance of this algorithm than the Least Square Support Vector Machine.

A GABOR FILTER BASED FINGERPRINT ENHANCEMENT SCHEME USING AVERAGE FREQUENCY

Fingerprint minutiae are prevalently used in fingerprint recognition systems. The extraction of fingerprint minutiae is heavily affected by the quality of fingerprint images. This leads to the incorporation of a fingerprint enhancement module in fingerprint recognition systems to make the system robust with respect to the quality of input fingerprint images. Most of existing enhancement methods suffer from two main kinds of defects: (1) time consuming and thus unusable in time critical applications; and (2) blocky and directional effects in the enhanced image. This paper proposes an improved fingerprint enhancement scheme based on the Gabor filter tuning its frequency to the average frequency of the input image and changing its shape from square to circle and dynamically adjusting the filters size based on the average frequency. This scheme can enhance the fingerprint image rapidly and overcome the blocky and directional effects and does improve the performance of minutiae detection.

Automatic cytoplasm and nuclei segmentation for color cervical smear image using an efficient gap-search MRF

Accurate and effective cervical smear image segmentation is required for automated cervical cell analysis systems. Thus, we proposed a novel superpixel-based Markov random field (MRF) segmentation framework to acquire the nucleus, cytoplasm and image background of cell images. We seek to classify color non-overlapping superpixel-patches on one image for image segmentation. This model describes the whole image as an undirected probabilistic graphical model and was developed using an automatic label-map mechanism for determining nuclear, cytoplasmic and background regions. A gap-search algorithm was designed to enhance the model efficiency. Data show that the algorithms of our framework provide better accuracy for both real-world and the public Herlev datasets. Furthermore, the proposed gap-search algorithm of this model is much more faster than pixel-based and superpixel-based algorithms.

Score Based Biometric Template Selection and Update

A biometric system usually contains two stages: registration and authentication. Most biometric systems capture multiple samples of the same biometric trait at registration. As a result, it is essential to select several samples as templates. This paper proposes two algorithms maximum match scores (MMS) and greedy maximum match scores (GMMS) based on match scores for template selection and update. The proposed algorithms need not involve the specific details about the biometric data. Therefore, they are more flexible and can be used in various biometric systems. The two algorithms are compared with Random and sMDIST on the database of FVC2006DB1A, and the experimental results show that the proposed approaches can improve the accuracy of biometric system efficiently. Based on the maximized score model, we propose two strategies: ONLINE and OFFLINE for templates update. And then we analyze the relationship between them. Preliminary experiments demonstrate that OFFLINE strategy gains better performance.

A Scalable Algorithm for Graph-Based Active Learning

In many learning tasks, to obtain labeled instances is hard due to heavy cost while unlabeled instances can be easily collected. Active learners can significantly reduce labeling cost by only selecting the most informative instances for labeling. Graph-based learning methods are popular in machine learning in recent years because of clear mathematical framework and strong performance with suitable models. However, they suffer heavy computation when the whole graph is in huge size. In this paper, we propose a scalable algorithm for graph-based active learning. The proposed method can be described as follows. In the beginning, a backbone graph is constructed instead of the whole graph. Then the instances in the backbone graph are chosen for labeling. Finally, the instances with the maximum expected information gain are sampled repeatedly based on the graph regularization model. The experiments show that the proposed method obtains smaller data utilization and average deficiency than other popular active learners on selected datasets from semi-supervised learning benchmarks.

Quality estimation of fingerprint image based on neural network

Quality estimation of fingerprint image can be used to control image quality at the enrollment stage of automatic recognition system and guide the enhancement of fingerprint image. This paper proposes a neural network based fingerprint image quality estimation method. It estimates the correctness of ridge orientation of each local image block using neural network and then computes the global image quality based on the local orientation correctness. The proposed method is used to guide the fingerprint enrollment and improves the accuracy of the automatic fingerprint recognition system.

Fingerprint alignment using special ridges

Fingerprint is one of the biometrics used to identify a person. Alignment is an important step in fingerprint recognition, affecting greatly the speed and accuracy of matching. Eight types of Special ridges are introduced to align two fingerprints. The ridge with maximum of sampled curvature is used as reference ridges for initial alignment. And corresponding special ridges paired by topology get aligned by their features. The alignment parameters of translation and rotation finally come from all aligned special ridge pairs. Experiments show that alignment using special ridges is fast and robust.

A composite fingerprint segmentation based on Log-Gabor filter and orientation reliability

A robust fingerprint segmentation technique using adaptive threshold after Log-Gabor filtering and orientation reliability is proposed. The Log-Gabor filter turns the non-ridge areas of fingerprint image dark, but makes the ridge areas brighter than non-ridge areas. The gray features of filtered image are robust in spite of different gray features in original fingerprint images. An adaptive threshold according to the histogram is robust to exclude non-ridge areas. And Orientation Reliability is defined on the orientations of blocks to discriminate the disordered ridge like areas. Then fusion of the two segmentations and post-processing are introduced. Experiments show that the proposed composite segmentation technique is effective and robust to different contrast areas and disordered ridge areas in one image.