Palm Vein Recognition: A Comparative Research Paper

The Authors of the paper promised fair compensation for proofreading, reducing similarity (from 24% of copy-paste work to 2% ) and editing of paper, But they ended up cheating with the amount by paying only 7 USD and asked me to "Be Happy" 

Palm Vein Recognition: A Comparative Research

Paper

Fahad Mira, Mohammed Saud Miraa

https://ieeexplore.ieee.org/author/37085905310

College of Computer and Information Technology, Department of computer engineering, the University of Bedfordshire 

University of Multi Media, Cyberjaya, Malaysia 

 Abstract—The main reasons the palm vein evolved into an authentication technique are that its outcomes are challenging to manipulate or abuse due to its location inside the palm. Protecting data from unapproved access and undesired individuals has become simple. Three papers that addressed the palm vein in various contexts were identified during this investigation. This paper examines the findings of a case study to look into the dominant palm vein and compares the results of three related studies. The data was gathered from the literature reviews and research papers. The results of this research may have clear significant implications for using multiple feature extraction algorithms to achieve high accuracy and obtain satisfactory results. It allows palm vein researchers to explore a more precise image pre-processing tool, identical studies using additional datasets, and other local vein recognition function algorithms.

Index Terms—Palm vein, Biometric Authentication, Biometric Security, Gabor Filter, Comparative Study

 

I.    INTRODUCTION

More protection tools have appeared, considering the rapid development of technology, as seen by advancements in the usage of new computing systems, mobile devices, and Internet networks, among other areas [1]. Such devices verify the user’s identity to efficiently and securely monitor the entry and to stop undesired users from obtaining the necessary data. The critical measures to assess and verify the consumer’s identity may be one of the most influential sciences in this field [2]. By presenting appropriate and considered proof that the stated requirements for accessing the data are met, the identity of the approved individual may be verified. Palm’s vein prevents unwanted access to sensitive information and maintains its integrity. It is a reliable method because a vein is located beneath the skin and unaffected by ageing [3]. Several filters, like the Gabor and Gaussian, can be applied to the veining process to extract the most relevant information. The safety community widely acclaims the palm vein pattern as a fantastic strategy [4], [6]. So, palm vein has widespread in the world. So, there is a need to provide multiple feature extraction algorithms to achieve high accuracy and obtain satisfactory results. Therefore, it is necessary first to review the literature that has been written on this subject before attempting to determine the degree and accuracy of the methods used in these studies.

 

Fahad Mira is with the College of Computer and Information Technology Department of computer engineering, University of Bedfordshire, e-mail: bart.simpson&homer.simpson@uspringfield.edu (Corresponding author).

II.    METHODOLOGY

The researcher used current literature to do a comparative review that may help to explore the different ways palm vein applies this research. The related literature was collected via research papers and studies. This research is comparative and descriptive.

 

A.    Statement of the problem

Given the importance of the palm vein, it was decided to conduct this study to identify the many types of palm veins. And compare it with other studies to develop recommendations that will help propose multiple feature extraction algorithms. This agrees with Abed [1], who affirms the potential of PVS as an identity verification method. For many reasons, it is important; the vein occurs within the human body; unlike other recognition techniques, it is impossible to alter patterns such as changing the vein location from one part to another.

 

B.    Research Questions:

1.   What are the differences between the three studies in terms of their methods of palm vein?

2.  What is the proposed method for a developed palm vein?

 

C.    Objectives of the research

This research tries to determine the dominant palm vein forms in the three studies and compares them to conclude with recommendations that may help propose palm veins by multiple feature extraction algorithms.

 

D.    Significance of the research

The present research highlights the importance of the best palm vein authentication system, which may help to propose multiple feature extraction algorithms to achieve high accuracy and obtain satisfactory results.

 

E.    Definition of the Terms

Palm Vein: A biometric authentication approach that uses patterns of the palm vein, a vascular representation of an individual’s palm that can be used as a pattern for personal information [1].

 

F.   Limitations of the study

This research only focuses on using Gabor filters to verify palm and wrist vein patterns and create a vein recognition model that combines characteristics from both. All of the results and data were culled from the existing literature.

III.    LITERATURE REVIEW

A.    Fusion-based vein palm recognition

The researchers proposed a four-step methodology to arrive at the findings: The histogram equation was used to improve the image and display its qualities in the first step of pre-processing, and the Gabor filter was employed in the second stage to extract those features; and in the third stage, the employment of two- discrete wavelet filters was recommended for the extraction of features. In the final step, data analysis or feature reduction is performed using PCA. Distances between components were ultimately determined using the Euclidean metric. The findings were reasonable because of the similarity ratio of 96.2 per cent. After several tests, these results were obtained, and the researchers got the best results using a GA filter with two-discrete wavelet transformation and PCA [5].

 

B.    Gabor filter wrist/palm vein pattern identification

The application of the Gabor filter to detect wrist and palm vein patterns is intriguing. Using wrist and palm vein patterns for identification has shown promise as a biometric method. Vein, in contrast to other types of identification techniques, remains within the human body, making it hard to change patterns by, for example, relocating the position of a vein from one region of the body to another. In this research, wrist and palm veins are used for identification and testing purposes, and the analysis is broken down into three sections: pre-processing, extraction of features, and recognition. During pre-processing, images are resized and “enhanced” with “CLAHE and 2-D Gaussian high pass filter,” and then features are extracted from the images with the help of Gabor filters. LDA and PCA are used to reduce the size of the characteristics package. To determine how similar images of veins were, the identification process employed the Euclidean distance. In the suggested work, the average palm vein CRR is 94.49 per cent, and the average vein wrist CRR is 92.33 per cent [1].

 

C.    Palm vein verification using the Gabor filter:

The advantages of palm vein over traditional biometrics (fingerprint, iris, and facial recognition) include its low falsifi- cation risk, high replication difficulty, and long-term reliability. Palm vein traits are proposed as a novel method for personal authentication in this research. The suggested method involves enhancing pictures of the palm vein and then extracting characteristics using a Gabor bank. After that, we apply  Fisher Discriminant Analysis (FDA) to reduce the number of dimensions in the functions’ vector space. This work uses the Nearest Neighbors technique for vein pattern verification. The suggested method’s EER is 0.2335 per cent [4].

 

 

 

 

 

 

Fig. 1. A comparison table between the studies

 

IV.   DISCUSSION

The benefits of the classical biometric palm vein are a small probability of falsification, replication complexity and stability. However, to answer question one, the first research study suggested a four-stage model to achieve the desired results; the following steps were taken: in the first stage, The image was enhanced, and the attributes were shown using the histogram equation.; the second stage utilised Gabor for the feature extraction and the third step recommended 2-discrete wavelet filters. Two-discrete wavelet filters are highly prized for their ability to analyse features and shrink feature spacing; principal component analysis (PCA) is employed in the last stage to compress data or features further. In the last phase, distances were determined using the Euclidean method. Since the calculated similarity was so near to the Euclidean distance, the findings were reasonable and convincing. The proposed model was tested on a dataset of palm vein images collected from 50 volunteers (two sets of images each for the right and left hand) over three sessions (four images each), separated by at least one week. The experiments aimed to establish the user’s identity and the approved person’s level of expertise. The first experimental result from the dataset for detecting functions is that vein vessel lighting was improved with histogram equalisation after pre-processing. In the second experiment, we use principal component analysis to determine which features best characterise each vein picture by minimising the natural characteristics of the photos. These features are based on Gabor fusion and the two-dimensional dwt coefficient.

As demonstrated in the previous section, many examples were utilised to implement the testing procedure. Initially, as individual features, the researcher offered Gabor coefficients, resulting in a right-hand precision of 85.5% and a left-hand precision of 84.5%. The accuracy was higher than in the initial experimental test on the collection and extraction of characteristics and on employing 2D-DWT as characteristics. After that, PCA is used to double-check the features from the prior approaches, as the improved accuracy and performance accuracy were both welcome developments. It was 96 percent, and in the end, the best feature extraction methods were those based on the fusion coefficient, which showed 96 per cent accuracy for the right palm vein and 96.2 percent for the left palm vein. Receiver operating characteristic (ROC) plots of the erroneous accept rate and false reject rate suggest an acceptance threshold of 4.7. [5].

In contrast, the second research used a three-step pre-processing pipeline that included resizing and “enhancement” image processing using “CLAHE and (2-D) Gaussian high pass filter,” feature extraction with Gabor filters, and detection of wrist and palm veins. Minimising the function set’s dimensions was the goal of the LDA and PCA analyses. Vein images were compared using Euclidean distance for identification. In the suggested work, the average palm vein CRR is 94.49 percent, whereas the average vein wrist CRR is 92.33 percent. For this study, we compiled a collection of 2400 images from 50 different students. The proposed work obtains a positive outcome and success proven by the experiment. Finally, using Matlab R2015a, the method has been introduced. The resulting rate for the “PCA”-based recognition system obtained positive values higher than that used for both palm and wrist “LDA” features reduction [1]. However, the third study, this research proposes a new approach to personal authentication focused on palm

Vein functionality. The photos of the palm vein are first enhanced in the proposed method; then, the characteristics are collected using the Gabor bank. Fisher Discriminated Analysis (FDA) is then used to reduce the vector dimension of the functions. For vein pattern verification, this work uses the Nearest Neighbors technique. The EER of the proposed approach is 0.2335 per cent. Histogram equalisation is used to enhance Palm vein visualisations. The convolution pictures are then employed as function vectors, and the augmented images are processed via a Gabor bank filter. The correct characteristics for verification are obtained via the FDA, and then the dimensional reduction is performed. Finally, using the Closest Neighbor classifier, palm vein verification was added. The images in the user database are 6000 pictures for 500 individuals [4].

Regarding the answer to question two, the researcher agrees with [2], [7] [8], [9]; therefore, based on three tended feature extraction techniques, the researcher suggests palm vein recognition. Researchers recommend contrasting three local invariant feature extraction techniques to determine the most useful for the palm vein identification system. Improve the Speed of a Scale-Invariant Feature Transform Affine-SIFT-based palm vein identification employing powerful characteristics. The photos are pre-processed by histogram equalisation, then local features are extracted using one of three techniques, and lastly, results are compared using the Euclidean distance. This database, together with the (PolyU) multispectral palm print database, is what the researcher believes has the potential to be very effective.

The key stages of the suggested approach to identifying vein images from the database are as follows:

1.    Image pre-processing: Histogram equalisation removes the Area of Interest from palm vein photographs, improving picture quality and contrast during the database’s selection phase. Images will be pre-processed and then utilised to extract neighbourhood characteristics.

2.    Extraction of local invariant features: Three algorithms will be implemented to extract SIFT characteristics, SURF and ASIFT functionality.

3.    Image match and recognition: Local feature points in two photos will be matched using an enhanced version of the ratio-based Euclidean distance similarity.

V.  CONCLUSIONS AND RECOMMENDATIONS

Palm vein detection has recently become the most used biometric identification method. Since the hand vein, along with the palm vein, the finger vein, and the back vein, is extremely distinctive and impossible to alter by action, and since it is also more convenient and simpler for data collection for a non-contact acquisition method, it is much more accurate than fingerprints or other biological characteristics.

The author suggested using the SIFT, SURF, and ASIFT local invariant feature algorithms for palm vein detection with the expectation of successful outcomes. Three algorithms will apply a histogram equalisation method to pre-process the gathered pictures and extract local features. If the experiment yields the desired results, all three should be able to correctly match and recognise a photo of a vein taken with the same hand. It’s really difficult to tell between images of different hands. Also, ASIFT offers the maximum accuracy, while the SURF algorithm provides the best synthetic efficiency.

There are several potential avenues for future studies, such as a more accurate picture pre-processing tool, comparable studies on more datasets, and alternative local vein recognition function algorithms.

ACKNOWLEDGMENT

The school of computer science technology at the University of Bedfordshire in the United Kingdom has allowed the author to conduct this research, and he would like to express his heartfelt gratitude for that.

 

REFERENCES

[1]    M. Abed and Hamzah, “Wrist and palm vein pattern recognition using Gabor filter,” Journal of AL-Qadisiyah for computer science and mathe- matics, vol. 9, no. 1, pp. 49–60, 2017.

[2]    M. Pan and W. Kang, “Palm Vein Recognition Based on Three Local In- variant Feature Extraction Algorithms,” in Biometric Recognition. CCBR 2011, S. Z., L. J., C. X., and T. T., Eds., vol. 7098. Springer.

[3]    P. Tahmasebi, A. Hezarkhani, and M. Mortazavi, “Application of discrim- inant analysis for alteration separation,” Australian Journal of Basic and Applied Sciences, vol. 4, no. 4, pp. 564–576, 2010.

[4]    Al-Juboori, ““Palm vein verification using Gabor filter,” International Journal of Computer Science Issues, vol. 10, no. 1, 2013.

[5]    Al-Hussain Hadi and Q. A, “Vein palm recognition model using fusion  of features,” Telkomnika, vol. 18, no. 6, pp. 2921–2927, 2020.

[6]    Nakisa and Bahareh, “Technology Acceptance  Model:  A  Case  Study of Palm Vein Authentication Technology,” IEEE Access,  vol.  10,  pp. 120 436–120 449, 2022.

[7]    N. Karennagari, K. Reddy, V. K. Gurrala, K. Srinivas, A. Peddi, and Y. P. Sai, “Infection Segmentation of Leaves Using Deep Learning techniques to enhance crop productivity in smart agriculture,” 2021 6th International Conference on Signal Processing Computing and Control (ISPCC), pp. 368–372, 2021.

[8]    W. Masaki, T. Endoh, M. Shiohara, and S. Sasaki, “Palm vein authen- tication technology and its applications,” Proceedings of the biometric consortium conference, pp. 19–21, 2005.

[9]    K. Srinivas and Kalyana, “Artificial Intelligence based Optimal Biometric Security System Using Palm Veins,” 2022 International Mobile and Embedded Technology Conference (MECON), pp. 2022–2022.