Step 2. The LBP histograms of the local windows are calculated and connected to obtain an eigenvector including values. Step 3. The LBP histogram of the near finger knuckle print is connected to form the final histogram feature as the extracted finger knuckle print feature, and its vector dimension is 3, The ROI images of four-finger knuckle prints under different light conditions are shown in Figure 8.
The histogram of two sets of ROI images is shown in Figure 9 a , and the blue histogram represents the images that are brighter, while the red histogram represents the images that are darker. Figure 9 b shows the LBP histogram characteristics of two different collected samples for comparison. It can be clearly seen from the details that there are minor differences among the LBP histograms of the same sample under different illumination, while those of different samples are much greater.
It further confirms that the illumination robustness of the LBP operator can be used as a feature extraction method for noncontact biological features, especially texture feature recognition. To verify the effect of different numbers of selected points and radius on the recognition accuracy rate and recognition time in the LBP operator, this paper had selected the various point numbers P and radii R of the round LBP operator for comparative experiments, and P , R were selected as 8, 1 , 8, 2 , 8, 3 , 16, 1 , 16, 2 , and 16, 3.
In addition, the ROI images of the four-finger near finger knuckle prints were collected for LBP feature extraction and recognition. The experimental results are displayed in Table 2. According to the comparison experiment results, when the number and radius of the LBP operators were 8 and 2, respectively, the recognition rate was up to The training time of ROI images was When there were 16 points, the recognition rate decreased, and the training time and test time also multiplied.
Obviously, the 8, 2 LBP operator had the best performance in recognition rate and running time. Global feature recognition extracts features with low dimensions, which are rough matching. The recognition speed is high but the recognition accuracy is low.
However, local feature recognition extracts features with high dimensions, which belongs to fine matching. The recognition rate is great, but the recognition speed is relatively low [ 20 — 24 ]. Therefore, in order to comprehensively utilize the advantages of the two features and improve the recognition rate and recognition speed of the recognition system, this paper adopted the serial fusion strategy to design a two-layer classifier.
The first-layer classifier mainly completed the global feature matching in the sample database to be tested. According to the matching result, the test sample library was initially screened, and all images in the database were sorted according to the similarity. The number of samples to be matched was decreased, and the samples with large differences were excluded. The second-layer classifier further identified the remaining samples according to the extracted local features and determined the recognition results.
The strategy can reduce the number of fine matching and improve the recognition speed substantially. The matching strategy of global and local fusion is shown in Figure The first-layer classifier was a minimum distance classifier.
In order to further improve the recognition speed, two fixed thresholds and were set in this paper. The Euclidean distance between the two images is. When the Euclidean distance between the image to be detected and the image in the database is smaller than , they belong to the same finger knuckle print.
When is greater than , they belong to different finger knuckle print. Therefore, the image of or will not be screened in the second-layer classifier.
After being screened by the first-layer classifier, the N image of is taken as input to the second-layer classifier. The selection of and can also influence the results. The larger the value, the lower the recognition speed of the second-layer classifier. The smaller the value, the more likely the image of the same type is excluded.
Therefore, it is necessary to find a balance point between the recognition rate and the recognition speed and select an appropriate distance threshold. The k-nearest neighbor algorithm was used in the second-layer classifier to calculate the chi-square distance between the LBP feature values. The recognition results were obtained according to the distance. In order to verify the effectiveness of the proposed fingerprint recognition algorithm combining global and local features, the following experiments were carried out.
Experiment 1. Intra- and interclass sample distance distribution density curve. Thresholds of and were determined according to the intra- and interclass sample distance distribution density curves. The collected sample database image was trained, and the Euclidean distances of all intra- and interclass images were calculated. The samples with the corresponding distances form the intra- and interclass sample distance distribution density curve, as shown in Figure The interclass sample indexing density corresponding to is 0.
Similarly, the intraclass sample distribution density corresponding to is 0. Therefore, the values of and are 2. Experiment 2. Comparison of different methods. The proposed serial matching fusion method was compared with PCA global feature recognition, LBP operator, and traditional fusion based on fractional fusion.
The recognition time and recognition rates are shown in Table 3. Global and local features of finger knuckle print represent different features of the image. Global features describe the overall feature and are more suitable for rough extraction. Local features describe image details and are suitable for detail recognition.
Therefore, combining the advantages of global features and local features in image recognition helps the recognition of finger knuckle print.
Global features with low dimensions were extracted through PCA, which had high recognition speed. According to the experiment results, the four-finger finger knuckle print has different classification weights and contribution rates for the recognition results.
The equal-weight distribution method of the existing research was improved. Local features were extracted through the IBP operator. The serial matching fusion strategy was adopted in the fusion of global and local features. The first-layer classifier mainly completed the global feature matching. The samples were initially screened according to the matching result, and all the images in the database were sorted based on the similarity degree.
The number of samples to be matched was reduced, and the samples with larger differences were excluded. The second-layer classifier further identified the remaining samples to be matched according to the extracted local features and determined the recognition result. The proposed algorithm is verified by the data set. In this way, the number of times of fine matching was decreased, and the recognition speed was improved substantially.
The image data that support the findings of this study are available from the corresponding author upon reasonable request. This is an open access article distributed under the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Article of the Year Award: Outstanding research contributions of , as selected by our Chief Editors.
Read the winning articles. Journal overview. Special Issues. Academic Editor: Deepak Kumar Jain. Received 18 Oct Revised 20 Nov Accepted 25 Nov Published 07 Jan Introduction Hand-based multibiometric recognition system occupies an important position in the field of biometric recognition.
Figure 1. Schematic diagram of finger knuckle print and ROI image: a finger knuckle print image and b ROI image of near finger knuckle print image. Figure 2. Some examples of image in the database: a index finger, b little finger, c middle finger, and d ring finger. Figure 3. Relation of cumulative contribution rate and the feature dimension.
Figure 4. Figure 5. Figure 6. Relationship between recognition rate and feature dimension: a weight ratio and b weight ratio Table 1. Figure 7. Figure 8. The ROI images of finger knuckle prints under different light conditions: a histogram characteristics of the same samples under light conditions and b histogram characteristics of two different collected samples.
Figure 9. Comparison of LBP histogram: a histogram characteristics of the same samples under light conditions and b histogram characteristics of two different collected samples. Table 2. Figure Schematic diagram of the serial matching strategy of global and local fusion. Intra- and interclass sample distance distribution density curves. Table 3. References L. Zhang, L. Zhang, D. Zhang, and H. Zhang, and Z. Aoyama, K. Ito, and T.
Zhang and H. Swati and M. Kumar and C. Gomaa, G. Salama, and I. Kumar and Z. Liu, L. Lv, and Y. Farah and M. Vidhyapriya and S. Wen-wen, L. Fu, and S. They are mostly used for multimodal biometrics are presented in section 3. Section people, while minimizing the chance of false positive 5 concludes the paper with some suggestions for or false negative in terms of authentication.
At the further investigations. As later demonstrated in this paper, there are many Photographs and fingerprints have been used as different biometrics that could be utilized in a given personal identification tools for many decades. More system. However, despite the wide variety of possible recently some more advanced biometrics tools have biometrics, the key high-level structure of a been developed, such as the case of measuring the biometrics system is the same [2].
These discriminating features could, for well as recognizing voices or faces, or the way people instance, be the relative locations of minutiae points walk. Multimodal biometric systems that infer extracted from a fingerprint [3] or an iris code from an personal identification based on multiple iris. Each sample's representation is referred to as a template.
Clearly, a system has to allow for the addition or one biometric before making an identification registration of new templates. This process is called decision. The templates are stored in a database and may be linked to particular information about the 3.
Numerous systems have been built that make use of one biometric for identification purposes. The Biometrics systems are usually used to accomplish biometric used can be especially suitable or one of two related objectives: identification or inappropriate for a given application depending on its verification [4]. Identification refers to the process of characteristic strengths and weaknesses.
We carry out trying to find a match between a query biometric a brief survey to contrast the different commonly used sample and one stored in the database.
For instance, to biometrics outlining the basics of how they work and obtain access to a restricted area, one may go through the motivations behind the use of each one. A corresponding template describing the discriminating features is built for the scanned 3.
Then the new template is compared to all those stored in the database. One is then granted or denied The iris begins to form in the third month of gestation, access depending on the existence of a database and its patterns become unchangeable by the age of template similar to the query template. Verification, two or three. Furthermore, the iris pattern depends on on the other hand, refers to the process of checking the initial environment of the embryo, hence making whether a query biometric sample belongs to a the iris a unique feature and highly distinct from one claimed identity.
For instance, one may have to enter person to another. In fact, even the left and the right an ID number or use a particular card, then have one's irises for the same person are not identical. The biometric system then has to only The iris is isolated and protected from external check whether the constructed query template is environment and it is impossible to surgically modify similar enough to the database template associated the iris without unacceptable risk to vision.
From these Additionally, its physiological response to light descriptions, it should be obvious that verification is provides one of several natural tests against artifice less computationally expensive and more robust than [6]. Thus, given its uniqueness and permanence, iris identification. However, identification is more recognition comes to be a popular biometric convenient, less obtrusive, and is the only option identification technology for personal identification.
After all, criminals are unlikely to put Iris recognition identifies a person by analyzing the forward their correct identities willingly.
The iris widely used metric to measure performance of recognition process has six main steps [7]: biometric system, so that systems could be compared, real-world performance can be estimated, and 1. Image acquisition, as shown in figure 1, is progress could be motivated. In biometrics, one of the major challenges of automated iris performance is ultimately based on the probability of recognition, because we need to capture a accepting impostor users, referred to as False high-quality image of the iris while Acceptance Rate FAR ; and the probability of remaining non-invasive to the human rejecting genuine users, referred to as False Rejection operator.
Rate FRR. Plotting the value of 1 - FRR against 2. Iris localization in which the edges of the iris FAR produces what is known as the Receiver and the pupil are detected to extract the iris Operating Curve, which could be used for a graphical region.
For a simple empirical measure, the Equal 3. Normalization of the size of the iris region to Error Rate EER is usually used in biometrics, which ensure consistency between eye images refers to the point at which FRR and FAR are equal despite the stretching of the iris induced by [5]. Unwrapping of the normalized iris region 3 Biometrics Techniques into a rectangular region. Various behavioural and physical attributes could be 5. Extraction of discrimination features in the used to identify an individual.
We will visit briefly iris pattern, so that a comparison between some of the most commonly used biometrics and templates can be done. Then, we will move on to 6. Encoding of iris features using wavelets to describe multimodal systems that consult more than construct the iris code to which input templates are compared in the matching step. Although exhibiting a number of strong points, later having not identified, alerted of, or caught any discussed above, iris recognition suffers from a few criminals, according to a spokesman for the Tampa problems as well.
For one thing, it is very difficult to Police Department [13]. Some of these approaches use the whole the head still and looking into the camera. Iris face as raw input, such as eigenfaces figure 2 and recognition is susceptible to poor image quality, with fisherfaces, which are based on principal component associated failure to enrol rates [8].
Besides the iris is analysis. Other approaches depend on extracting and located behind a curved, wet, and reflecting surface matching certain features from the face, such the and obscured by eyelashes, lenses, and reflections. It mouth and eyes. Lastly, some approaches are a mix of is partially occluded by eyelids. Also, it deforms non- the two using data from the whole face as well as elastically as pupil changes size [9]. All these reasons specific features to carry out the recognition [14].
Figure 2: Images generated by Eigenfaces approach. Figure 1: Iris scanner used to identify Baghdadi city [15] council members [10] 3. The uniqueness of friction ridges tells us that by many organizations, such as universities, no two fingers or palm prints are ever exactly alike; government agencies, and banks, although the even two recorded successive impressions from the recognition is usually carried out by a human. Many same finger are no identical [17].
Additionally, facial images of a person can usually be under some threshold scoring rules. Thus, it is no surprise that systems is that small injuries and burns highly affect facial recognition is a key part of the DARPA-funded the fingerprint.
In fact, injury, whether temporary or Human ID at a Distance Project aimed at developing permanent, can interfere with the scanning process. Despite the aforementioned advantages of using facial Something as simple as a burn to the identifying recognition, it may perform very poorly when finger could make the fingerprint identification deployed in the real world, especially for recognition process fail [18].
Also add to this that the type of at a distance. A facial recognition system deployed in work that a person performs can also affect and Logan International Airport to detect terrorists failed sometimes damage some of fingerprint ridges as well. This voice authentication process is based on On the other hand, DNA techniques are not able to an analysis of the vibrations created in the human distinguish between monozygotic twins which are vocal tract.
DNA faces several other challenges as well; everyone has a unique vocal tract in shape and size. This is why people are fairly choose [20]. However, Human voice is generated hostile to DNA usage [23].
And hence, any small 3. Use of hand used to hack into a system. A DNA fingerprint is robust and is impossible Similar to other technologies, hand geometry's ease of to be changed by surgery or any other known use and acceptably amongst users does come at a cost. It is widely used in the diagnosis of For one thing, hand geometry is not especially disorders, paternity tests, criminal identification.
Thus, it is most suitable for purposes of people with the same DNA profile is one to billion, verification rather than identification.
Additionally, making DNA testing at a very high degree of hand geometry may not be an ideal biometric to use if accuracy. Hand geometry recognition can be done based on the 3. Most hand-geometry systems determine the unreadable. Besides, intrapersonal variations and different parts of the hand based on pins between differences make the analysis of these signatures as fingers, restricting the position in which one can place complete images and not as letters and words one's hand [26].
Figure 5 shows one of such hand important and unique [32]. That is why signatures geometry recognition systems. There are two different types of signature verification: offline and online. The former takes only a signature's image and analyses it, and then compares it to the stored template to measure similarity figure 6. The latter goes a step further and various features that are determined by the signing method are evaluated.
Such features include the number of strokes used, the amount of pressure at a given point, and the writing speed. Online signature verification is more robust against forgery as it requires the forger to not only copy the signature's shape, but also copy the way it is written.
However, online techniques cannot be applied for verifying signatures on documents or bank cheques [33]. Figure 5: Hand Geometry Recognition system [27] 3. This ability to identify a person from a distance is gait's biggest asset, because it helps in building non-obtrusive authentications systems as well as helping to meet today's heightened security concerns.
Alongside facial recognition, the Human ID at a Distance Project uses gait to help identify terrorists from a distance [11]. Figure 6: Signature Verification Process [34] Additionally, it is difficult for one to deliberately copy someone else's way of walking [28]. As a matter of fact, even successive signatures by the same person can be significantly Most of current approaches to gait recognition are different.
Given video input of one's walk, a corresponding silhouette is formed, which is analyzed to produce a gait signature. The 3. Also, different hidden Markov models and eigen analysis [30].
Recently, a very different type of approach to gait Thus, conditions that may cause one biometric to fail recognition has emerged that relies on having a may have no influence on another. For instance, if physical device, such as an accelerometer, attached to one's voice had changed because of a cold, speech one's physical body to collect data about one's gait.
Alternatively, if one's finger incurred an injury, fingerprint identification may become less reliable than that of voice identification. The Borda count is a winner election method Figure 7: Multimodal Biometrics [35] in which voters rank candidates in order of preference. Candidates will receive a certain number Many approaches exist for integrating multiple of points depending on the position they hold. Finally, modalities. Fusion approaches can be distinguished in the winner of an election is the candidate with highest two ways.
First of all, fusion could be carried out at number of points. Modalities could be combined at the feature level, the matching level, or the decision level. One problem that appears with decision level fusion is Secondly, fusion could be based on rules or based on the possibility of having a tie.
Therefore it is machine-learning approaches [36]. Rule-based necessary to have more classifiers than classes. But for classifier, Fisher's linear discriminant, decision trees, identification cases, it is not practical and even and multilayer perceptron [38]. First, we categorize and present specific Euclidean distance. Feature Fusion produced best application examples based on the primary benefit of results when modalities are related e.
LDA-Red, biometrics that they make use of. Of course, these applications have Quadric-Line-Quadric function, which tries to sprung up because of the benefits biometrics provides separate the genuine and impostor score distributions as an identification solution.
Depending on the [37]. Then it may use one or more of following application, a particular benefit is usually emphasized, approaches for the actual classification [37]: such as security, convenience or privacy. The uniqueness of biometric signals and the difficulty in forging them makes biometrics an attractive 4.
This technology allows E-passport is an advanced smart card that holds for quick and reliable identification of citizens [41]. This card will substitute both the Bioscrypt's V-Smart authentication system to provide national identity card and the conventional passport access control to high-security areas of its terminals.
0コメント