CHAPTER ONE
INTRODUCTION
1.1. Background
The term “identification” means the act or process of establishing the identity of or recognising, the treating of a thing as identical with another, the act or process of recognising or establishing as being a particular person, but also the act or process of making representing to be, or regarding or treating as the same or identical.
Computerized human identity is one of the most essential and difficult tasks to meet developing call for stringent protection. The usage of physiological and/or behavioral characteristics of people, i.e., biometrics, has been drastically employed within the identity of criminals and matured as an essential tool for law enforcement departments. The biometrics-primarily based automatic human identity is now rather popular in a extensive range of civilian packages and has an end up effective alternative to traditional (password or token) identification systems. Human fingers are easier to offer for imaging and may screen a diffusion of statistics. Therefore, palmprint studies has invited a number of attention for civilian and forensic usage. But, like a number of the popular biometrics (e.g., fingerprint, iris, face), the palmprint biometric is likewise at risk of sensor level spoof assaults. Far flung imaging the usage of a excessive-resolution digicam may be hired to reveal important palmprint information for feasible spoof assaults and impersonation. consequently, extrinsic biometric capabilities are predicted to be greater vulnerable for spoofing with mild efforts. In precis, the blessings of clean accessibility of those extrinsic biometric tendencies additionally generate a few concerns on privateness and security. on the other hand, intrinsic biometrics characteristics (e.g., DNA, vessel systems) require greater hard efforts to accumulate with out the knowledge of an person and, consequently, extra tough to forge. but, in civilian applications it is also crucial for a biometrics trait to ensure excessive collectability at the same time as the consumer interacts with the biometrics tool. on this context, palm-vein popularity has emerged as a promising opportunity for non-public identification. It has the benefit of the high agility.
Biometrics is authentication using biological data. It is a powerful method for authenticating. The general purpose of biometry is to distinguish people from each other by using features that cannot be copied or imitated. There is less risk than other methods because it is not possible for people to change, lose, and forget their physical properties. The use of these features, defined as biometric metrics, in ciphers is based on an international standard established by INCITS (International Committee for Information Technology).
In recent years, many people have been able to distinguish amount of work has been done. Some of the patterns studied are characters, symbols, pictures, sound waves, electrocardiograms. Usually complex difficult to interpret due to calculations or human evaluations of overload problems are used in computerized identification. The path maps to the template. In this case, a template for each pattern class the set of templates is stored in memory in the form of a database. Unknown class of each class template. The classification is based on a previously determined mapping criterion or similarity criteria. Compare the pattern with the complete pattern it is faster to compare some features rather than the more accurate result most of the time. For this reason, pattern recognition process, feature extraction and classification examined in two separate phases.
In Picture 1.2 future extraction, makes some measurements on the pattern and turns the results into a feature vector. This feature may vary considerably depending on the nature of the problem. Also, the importance ratings and costs of the features may be different. For this reason, properties should be selected to distinguish the classes from each other and to achieve lower costs.
Features are different for every pattern recognition problem.
Based on the properties extracted in the classification stage, it is decided to which class the given object belongs to. Although the feature extraction does not differ according to the pattern recognition problem, the classifiers are collected in specific categories[6].
Template mapping is the most common classification method. In this method, each pixel of the view is used as a feature. The classification is done by comparing the input image to all the class templates. The comparison results in a similarity measure between the input information and the template, with the template, the pixel-based equivalence of the input image increases the degree of similarity, while the corresponding of the corresponding pixels reduces the similarity. After all the templates are compared, the class of the template giving the most similarity grade is selected. Structural classification techniques use structural features and decision rules to classify patterns. For example; line types in characters, holes and slopes are structural properties. Rule-based classification is performed by using these extracted features. Many pattern recognition systems are based on mathematical bases to reduce misclassification. These systems are pixel-based and use structural features. Examples include Gabor features, contour properties, gradient properties, and histograms. As a classifier, classifiers including discriminant function classifiers, Bayesian classifiers and artificial neural networks can be used[1].
In its simplest terms, it is a necessary tool to process darkness, manipulate images, and two important input-output niches are demanded image digitization and imaging devices. Due to the inherent nature of these devices, images do not create a direct source for computer analysis. Since computers work with numeric values rather than with image data, the image is transformed into a numeric format before processing begins. Picture-1.1 shows how a numbered array of numbers can represent a material image. The material image is divided into small regions called “shape elements” or “pixels”. The rectangular mesh cage device, which is the most comprehensive subdivision scheme, is also shown in Picture-1.1 In the digital image, the value placed on each pixel is given the brightness of that spot.
The conversion process is called numerical conversion. This situation is completely transferred to a diagram in Picture 1.2 The brightness of each pixel is used as examples and numerically. This part of the operation shows the brightness or darkness of each pixel in that place. When this process is applied to all pixels, the image is displayed in a rectangular shape. Each pixel has a full place or a trace (number of lines and columns), and at the same time has a full value called gray level. This sequence of numeric data is now available for processing on a computer. Picture 1.3 shows the numerical state of a continuous view.
1.2. Human Identity
Human beings cannot live without systems of meaning. Our primary impulse is the impulse to find and create meaning. But just as important, human beings cannot exist without an identity and often human identity is tied closely to the systems of meaning that people create. The systems of meaning are how people express their identity.
There are many elements that shape identity- family, community, ethnicity, nationality, religion, philosophy, science and occupation. For much of history, human identity has been oriented to small bands of extended families with belief systems that validated that lifestyle. With the movement toward domestication and state formation, along with the larger communities of such states, the boundaries of human identity were widened. But the small band mentality has persisted over subsequent millennia and is still evident even within modern states in the form of ethnic divisions, religious differences, occupation, social status, and even in the form of organizational membership[1].
The presence and origin of the small band mentality can be explained in terms of the inherited animal brain and its primitive drives. Animal life from the earliest time developed an existence of small groups of extended family members. This existence was shaped by the base drives to separate from others, to exclude outsiders, and to dominate or destroy them as competitors for resources. This in-group thinking and response was hardwired into the animal brain which has continued to influence the human brain. Unfortunately, small band mentality has long had a powerful influence on the creation of systems of meaning and the creation of human identity. People have long identified themselves in terms of some localized ethnic group, religion or nation in opposition to others who are not members of their group. This has led to the exclusion of outsiders and crusades to dominate or destroy them as enemies.
More recent discovery on human origins and development confirms the early Greek intuition. We now know that we are all descendent of a common hominid ancestor (the East Africa origin hypothesis). Race is now viewed as a human construct with little if any basis in real biology. So-called racial differences amount to nothing of any real distinction in biology. One scientist has even said that, genetically, racial features are of no more importance than sunburn.
This information points to the fact that we are all descendent of Africans. In the great migrations out of Africa some early hominids moved to Europe and endured millennia of sunlight deprivation and this led to a redistribution of melanin in their skin. They still possessed the same amount of melanin as that of darker skinned people but it was not as visible.
All this is to say that the human race is indeed one family. And modern human identity and meaning must be widened to include this fact. The small band mentality of our past which focuses human identity on some limited subset of the human race has always led to the creation of division, barriers, opposition and conflict between people. It is an animal view of human identity.
But we are no longer animals. We are now human and we need to overcome the animal tendency to separate from others, to exclude them, and to view them as outsiders or enemies to be dominated or destroyed.
It is also useful to note here how tightly many people tie their identity to the system of meaning that they adopt (their belief system or viewpoint). Consequently, any challenge to their system of meaning will produce an aggressive defensive reaction. The system may contain outdated ideas that ought to be challenged and discarded but because it comprises the identity of those who hold it, they will view any challenge as an attack on their very selves and this produces the survival response or reaction. Attacks on the self (self-identity) are viewed as attacks on personal survival and will evoke the aggressive animal defense. In this reaction we see the amygdala overruling the cortex.
This defensive reaction as an attempt to protect the self helps explain in part why people continue to hold on to outdated ideas and systems of belief/meaning. The ideas may not make rational sense to more objective outside viewers but to those who hold them, they make sense in terms of the dominant themes of their overall system.
It is true that we can’t live without meaning or identity. And our identity is often defined by our systems of meaning. This tendency to tie our identity too tightly to our systems of meaning calls for a caution: Human meaning and identity should not be placed in an object- a system of meaning, an ideology, an occupation, a state, a movement, ethnicity or some organization. Our identity and our search for meaning should be focused on the process of becoming human. This orients us to ongoing development and advance. We then remain open to make changes as new information comes along. It’s about the human self as dynamic process, not rigid and unchanging object.
So from our point of view, identity is used to mean the condition of being a specified person or the condition of being oneself and not another. It clusters with the terms personality and individualism, and less fashionably, “soul”.
Figure 1.1: Human Identity by face
1.3. Palm Print
Handprint recognition supplies fingerprint matching algorithms for nature: The two biometric systems are based on personal information, which is represented by the effects seen on the lines. Statistical analysis by FBI officials reflects the fact that handprint identification is a biometric system that is complementary to more popular fingerprint recognition systems. The findings of these studies show that 70% of traces left behind by criminals in crime scenes are from fingerprints and 30% are from palms. Because of the lack of processing capabilities and lack of live-scanning technologies, palm print recognition algorithms work more slowly when automated compared to fingerprint recognition algorithms. Since 1994, there has been a growing interest in systems that use fingerprint and palm print identification together. The palm print identification is based on massive information found on the friction ridge as on the fingerprint. The palm print, or fingerprint, consists of the dark lines representing the high and pointed portions of the lines of evidence, which are in sequence, and white lines representing the valleys between these lines of relief. The palm print recognition technology uses some of these characteristics.
Algorithms used for palm print detection and verification are similar to algorithms used in finger print recognition. These algorithms are basically based on correlation, based on feature points (minutiae) and based on ridges (ridges). Correlation-based matching involves two handfuls of images taken together to find corresponding lines in two images; feature-based matching is based on the determination of the location, orientation and orientation information of specific feature points in the palm image and comparison of these information. The line-based matching technique uses the geometric characteristics of the lines as well as the texture analysis, in addition to feature point analysis, while classifying the palm print.
Correlation-based algorithms work faster than other types of techniques, but they have less tolerance to distortions and rotation variances in the image. Algorithms based on feature points require high quality imagery and do not benefit from the textural or visual qualities of the player. Finally, line-based algorithms require a high-resolution sensor to produce good-quality images, as well as distinctive features of line characteristics that are significantly less than feature points. The positive and negative aspects of these techniques also apply to fingerprinting.
William James Herschel, the son of John Herschel, was an astronomer. His father asked him to choose a career other than astronomy, so he joined the East India Company, and in 1853 was posted to Bengal. Following the Indian Mutiny of 1858, Herschel became a member of the Indian Civil Service, and was posted to Jungipoor.
In 1858 he made a contract with Mr. Konai, he was a local man, for the construction of road building materials. To prevent Konai from rejecting his signature later, Herschel suppressed his handprint in his documentary figure 2.2 shows that Mr. Konai’s palmprints. Herschel continued to experiment with hand prints, soon he realized that it was necessary to use fingers. He collected prints from his friends and family, and the result was that one’s fingerprints did not change over time. The Governor of Bengal suggested that fingerprints should be used on legal documents to prevent impersonation and refusal of contracts, but this proposal was not addressed. [1]
Now we are using palm prints and fingerprints to investigate the criminal cases, for example in a crime scene we found some palmprints and fingerprints on a subject in the crime scene then we collect this print after that we compare the prints with the people who committed crime before. Also in government’s documents we use it like a sign of a person, and in health applications, so as we see there are many areas to use the palm print images.
Figure 1.2: Palm Print
1.3.1. Palm Print Features
Palm print has stable and rich line features, three types of line patterns are visible on the palm. They are principal lines, wrinkles, and ridges. Principal lines are the longest, and widest lines on the palm. The principal lines indicate the most distinguishing direction features on the palm. Most people have three principal lines, which are named as the heart line, head line, and life line. Wrinkles are regarded as the thinner and more irregular line patterns. The wrinkles, especially the pronounced wrinkles around the principal lines, can also contribute for the discriminability of the palm print. On the other hand, ridges are the fine line texture distributed throughout the palmar surface. The ridge feature is less useful for discriminating individual as they cannot be perceived under poor imaging source. Figure 4 shows the palm lines.
1.3.2. The importance of palm print identification
- Every person’s palm print is unique, so palm print identification is a perfect form of authentication.
- The palm print recognition system has a high level of security because it is impossible to steal.
- Palm print is used in many industries such as healthcare, aviation, education, construction and banking. Thus, palm print identification is a user-friendly system.
- The size of the palm print recognition system is small and portable.
- Palm print recognition system is hygienic due to contactless use.
1.4. Biometric Features
Physiological features include DNA, iris, finger prints, palm prints, and facial features, while behavioral features include mimics, signature, and sound. When measuring physiological / behavioral characteristics, such factors as age, health or mental status of the person should be eliminated from the measurement. The existing identification systems are not sufficient, the conventional methods based on the use of a personal identification number (PIN) together with user name and plastic cards are both inconvenient and unsafe. The ideal biometric based person recognition system should identify the identity of the individual uniquely or verify identification within the database accurately, reliably and most efficiently. For this reason, the system should be able to cope with problems such as inlet deterioration, environmental factors and signal mixtures, and should not change over time and be easily applicable. The most commonly used biometric feature is the most reliable iris scan while fingerprinting.
In this Project, we will work on the palm print recognition system, which is one of the physical features. Palm tracking has advantages over other biometric features. The required images are collected with a low-cost operation and the image does not cause any deterioration; False Accept Rate and False Reject Rate are reasonable values. Incorrect acceptance and rejection rates for a system are part of the total number of identification attempts for total false acceptance / rejection..
CHAPTER TWO
LITERATURE REVIEW
2.1.Background
In this chapter, I will talk about the approaches of palm print. Individual validation utilizing palmprint pictures has gotten extensive consideration the most recent 5 years and various methodologies have been proposed in the writing. The accessible methodologies for palmprint validation can be isolated into three classifications essentially on the premise of separated highlights; (I) surface based methodologies (ii) line-based methodologies, and (iii) appearance-based methodologies. The depiction of these methodologies is past the extent of this paper. However a synopsis of these methodologies with the run of the mill references can be found in Table 1. Analysts have indicated promising outcomes on inked pictures, pictures procured specifically from the scanner and pictures procured from advanced camera utilizing compelled pegged setup. However endeavors are as yet required to enhance the execution of unconstrained pictures gained from sans peg setup. Accordingly this paper uses such pictures to research the execution change. An outline of earlier work in Table 1 demonstrates that there has not been any endeavor to explore the palmprint confirmation utilizing its numerous portrayals [2].
A few coordinating score level combination methodologies for joining different biometric modalities have been displayed in the writing. It has been demonstrated that the execution of various combination methodologies is unique. Be that as it may, there has not been any endeavor to consolidate the choices of different score level combination techniques to accomplish execution change. The association of rest of this paper is as per the following; Section 2 depicted the square outline of the proposed framework. This area likewise points of interest include extraction strategies utilized in the tests. Area 3 subtle elements the coordinating paradigm and the proposed combination technique. Analysis comes about and their exchange show up in Section 4. At last the finishes of this work are compressed in Section.
2.2. Proposed Systems
Unlike previous work, we propose an alternative approach to palmprint authentication by the simultaneous use of different palmprint representations with the best pair of fixed combination rules. The block diagram of the proposed method for palmprint authentication using the combination of multiple features is shown in Fig. 1. The hand image from every user is acquired from the digital camera. These images are used to extract region of interest, i.e. palmprint, using the method detailed in Ref. [5]. Each of these images is further used to extract texture-, line- and appearance-based features using Gabor filters, Line detectors, and principal component analysis (PCA) respectively. These features are matched with their respective template features stored during the training stage. Three matching scores from these three classifiers are combined using fusion mechanism and a combined matching score is obtained, which is used to generate a class label, i.e., genuine or imposter, for each of the user. The experiments were also performed to investigate the performance of decision level fusion using individual decisions of three classifiers. However, the best experimental results were obtained with the proposed fusion strategy which is detailed in Section 4.
Figure 2.1: Block diagram for personal authentication using palmprint
2.2.1. Gabor Features
The surface highlights separated utilizing Gabor channels have been effectively utilized in unique mark characterization, penmanship acknowledgment and as of late in palmprint. In spatial area, an even-symmetric Gabor channel is a Gaussian capacity tweaked by a situated cosine work [3]. The motivation reaction of even-symmetric Gabor channel in 2-D plane has the accompanying general frame:
In this work, the parameters of Gabor filters were empirically determined for the acquired palmprint images. If we filter the image with Gabor filter, we get:
where ‘∗’ indicates discrete convolution and the Gabor channel cover is of size W×W. Accordingly every palmprint picture is sifted with a bank of six Gabor channels to create six separated pictures. Each of the separated pictures emphasizes the unmistakable palmprint lines and wrinkles in relating bearing i.e., while lessening foundation clamor and structures in different headings. The segments of palmprint wrinkles and lines in six unique ways are caught by each of these channels. Every one of these pictures separated picture is partitioned into a few covering squares of same size. The component vector from every one of the six sifted pictures is framed by figuring the standard deviation in every one of these covering squares. This component vector is utilized to remarkably speak to the palmprint picture and assess the execution [3].
Figure 2.2: Partial-domain representation
2.2.2. Extraction of line features
Palmprint distinguishing proof utilizing line highlights has been accounted for to be effective and offers high precision. The extraction of line highlights utilized as a part of our tests is same as definite in [4]. Four directional line indicators are utilized to test the palmprint wrinkles and lines arranged at each of the four bearings, i.e. 0 ◦, 45◦, 90◦ and 135 ◦. The spatial degree of these covers was experimentally settled as 9 × 9. The resultant four pictures are consolidated by voting of dark level extent from relating pixel position. The joined picture speaks to the consolidated directional guide of palm-lines and wrinkles in the palmprint picture. This picture is additionally partitioned into a few covering square pieces. The standard deviation of dim level in every one of the covering squares is utilized to shape the element vector for each palmprint picture[2].
The proposed palmprint verification technique was examined on a dataset of 100 clients. This informational index comprises of 1000 pictures, 10 pictures for each client, which were procured from advanced camera utilizing unconstrained sans peg setup in indoor condition. Fig. 5 indicates normal procurement of a hand picture utilizing the advanced camera with live criticism. The hand pictures were gathered over a time of 3 months from the clients in the age group of 16– 50 years. The hand pictures were gathered in two sessions from the volunteers, which were not very helpful. Amid picture procurement, the clients were just asked for to ensure that (I) their fingers don’t touch each other and (ii) the vast majority of their hand (rear) touches the imaging table. The mechanized division of locale of intrigue, i.e. palmprint, was accomplished by the technique itemized in Ref. [5]. Hence the palmprint picture of 300 × 300 pixels were acquired and utilized in our analyses. Every one of the obtained pictures was further histogram evened out.
2.2.3. Extraction of PCA features
The data substance of palmprint picture additionally comprises of certain nearby and worldwide highlights that can be utilized for distinguishing proof. This data can be extricated by enrolling the varieties in an ensemble of palmprint pictures, autonomous of any judgment of palmprint lines or wrinkles. Each N × N pixel palmprint picture is spoken to by a vector of 1 × N2 measurement utilizing line requesting. The accessible set of K preparing vectors is subjected to PCA which creates an arrangement of orthonormal vectors that can ideally speak to the data in the preparation dataset. The covariance framework of standardized vectors j can be gotten as takes after [2]:
2.3. Matching criterion
The grouping of separated component vectors utilizing each of three techniques is accomplished by closest neighbor (NN) classifier. The NN classifier is a basic nonparametric classifier which figures the base separation between the include vector of obscure example g and that of for gm in the mth class [5]:
where and individually speak to the nth part of highlight vector of obscure example and that of mth class. Every one of the three capabilities got from the three unique palmprint portrayals were explored different avenues regarding each of the over three separation measures (8)– (10). The separation measure that accomplished best execution was at long last chosen for the order of capabilities from the comparing palmprint portrayal.
The combination system goes for enhancing the joined grouping execution than that from single palmprint portrayal alone. There are three general techniques for joining classifiers; at include level, at score level and at choice level. Because of the expansive and shifting measurement of include vectors, the combination approach at highlight level has not been considered in this work. An outline [20] of utilized approaches for multimodal combination recommends that the score level combination of capabilities has been the most well-known approach for combination and has appeared to offer noteworthy change in execution. The objective of assessing different score level combination techniques is to create most ideal execution in palmprint verification utilizing given arrangement of pictures. Let LGabor(g, gm),LLine(g, gm) and LPCA(g, gm) signify the coordinating separation delivered by Gabor, Line and PCA classifiers separately. The joined coordinating score LC(g, gm) utilizing the outstanding settled tenets can be acquired
Figure 2.3: Combination of Gabor, Line, and PCA
I is the chosen consolidating guideline, i.e. I speaks to most extreme, aggregate, item or least lead (shortened as MAX, SUM, PROD and MIN separately), assessed in this work. One of the deficiencies of settled guidelines is the supposition that individual classifiers are autonomous. This supposition might be poor, particularly for the Gabor and Line based highlights. In this way SUM manage can be better option for uniting coordinating scores while joining Gabor furthermore, Line highlights. These merged coordinating scores can be additionally joined with PCA coordinating scores utilizing PROD manage (Fig. 3) as the PROD administer is evaluated to perform better on the supposition of free information portrayal [17]. The individual choices from the three palmprint portrayals were additionally joined (dominant part voting) to look at the execution change. The exhibitions of different score level combination techniques are unique. Along these lines the execution from straightforward crossover combination methodology that joins choices of different settled score level combination plans, as appeared in Fig. 4, was likewise examined in this work. of utilizing settled blend manages, the coordinating scores from the preparation set can likewise be utilized to adjust a classifier for two class, i.e. real and fraud, characterization. Hence the consolidated arrangement of three coordinating scores utilizing feed forward neural system (FFN) and bolster vector machine
(SVM) classifier has additionally been explored [5].
Figure 2.4: Hybrid fusion scheme
2.4. Image Acquisition & Alignment
Our picture procurement setup is innately basic and does not utilize any exceptional enlightenment (as in [3]) nor does it utilize any pegs to make any bother clients (as in [20]). The Olympus C-3020 computerized camera (1280 ‘ 960 pixels) was utilized to get the hand pictures as appeared in figure 2. The clients were just asked for to ensure that (I) theirs.
2.4.1. Extraction of hand geometry images
Every one of the gained pictures should be adjusted a favored way in order to catch the same highlights for coordinating. The picture thresholding activity is utilized to acquire a parallel hand-shape picture. The edge esteem is consequently figured utilizing Otsu’s technique [25]. Since the picture foundation is steady (dark), the limit esteem can be processed once and utilized in this manner for different pictures. The binarized state of the hand can be approximated by an oval. The parameters of the best-fitting circle, for a given double hand shape, is registered utilizing the minutes [26]. The introduction of the binarized hand picture is approximated by the significant pivot of the oval and the required edge of turn is the diffe rence amongst typical and the introduction of picture [6]. As appeared in figure 3, the binarized picture is turned and utilized for registering the hand geometry highlights. The assessed introduction of binarized picture is additionally used to turn dim level hand picture, from which the palmprint picture is extricated as nitty gritty in the following subsection.
Figure 2.5: Extraction of two biometric modalities from the hand image
2.4.2. Extraction of palmprint images
Each binarized hand-shape picture is subjected to morphological disintegration, with a known double SE, to register the locale of intrigue, i.e., the palmprint. Give R a chance to be the arrangement of non-zero pixels in a given parallel picture and SE be the arrangement of non-zero pixels, i.e., organizing component. The morphological disintegration is characterized as
where SE g signifies the organizing component with its reference point moved by g pixels. A square organizing component (SE) is utilized to test the composite binarized picture. The inside of twofold hand picture after disintegration, i.e., the focal point of rectangle that can encase the deposit is resolved. This inside directions are utilized to remove a square palmprint locale of settled measure as appeared in figure 3.
2.4.3. Extraction of hand geometry features
The two fold image‡ as appeared in figure 3(c), is utilized to register critical hand geometry highlights. An aggregate of 16 hand geometry highlights were utilized (figure 5); 4 finger lengths, 8 finger widths (2 widths for each finger), palm width, palm length, hand region, and hand length. In this way, the hand geometry of each hand picture is portrayed by a component vector of length 1×16. The various bits of confirmations can be consolidated by various data combination techniques that have been proposed in the writing. With regards to biometrics, three levels of data combination plans have been recommended; (I) combination at portrayal level, where the element vectors of different biometric are connected to shape a joined highlight vector, (ii) combination at choice level, where the choice scores of different biometric framework are consolidated to produce a ultimate conclusion score, and (iii) combination at dynamic level, where numerous choice from different biometric frameworks are combined. The first two combination plans are more important for a bimodal biometric framework and were considered in this work.
Figure 2.6: Hand geometry feature extraction
CHAPTER FOUR
IMPLEMENTATION
4.1. Background
In this project …. .
CHAPTER FIVE
CONCLUSION
5.1. Background
References
- A. Kumar, D. C. Wong, H. C. Shen, and A. K. Jain, “Personal verification using palmprint and hand geometry biometric,” in International Conference on Audio-and Video-Based Biometric Person Authentication, 2013, pp. 668-678.
- A. Kumar and D. Zhang, “Personal authentication using multiple palmprint representation,” Pattern Recognition, vol. 38, pp. 1695-1704, 2010.
- S. Pathania, “Palm Print: A Biometric for Human Identification,” 2016.
- J. Kodl and M. Lokay, “Human Identity, Human Identification and Human Security,” in Proceedings of the Conference on Security and Protection of Information, Idet Brno, Czech Republic, 2010, pp. 129-138.
- S. Sumathi and R. R. Hemamalini, “Person identification using palm print features with an efficient method of DWT,” in Global Trends in Information Systems and Software Applications, ed: Springer, 2012, pp. 337-346.
- H. J. Asghar, J. Pieprzyk, and H. Wang, “A new human identification protocol and Coppersmith’s baby-step giant-step algorithm,” in International Conference on Applied Cryptography and Network Security, 2010, pp. 349-366.
2018-4-25-1524696117