http://annalsofrscb.ro 17108
DESIGN AND DEVELOPMENT OF MULTIMODAL CYBER FORENSIC SYSTEM
A.Revathi 1*, Dr.Savadam Balaji 2, Renuka Kondabala 3, S V Vandana A Somayajulu 4
1Assistant Professor, VNR Vignana Jyothi Institute of Engineering and Technology, Hyderabad, Telangana
& Research Scholar, Department of CSE, Koneru Lakhmaiah Education Foundation, Hyderabad, India
2Professor, CSE Department, KL University, Hyderabad, India,
3,4
Assistant Professor, VNR Vignana Jyothi Institute of Engineering and Technology, Hyderabad, Telangana
&Research Scholar, Department of CSE, Koneru Lakhmaiah Education Foundation, Hyderabad, India
* [email protected], [email protected], [email protected], [email protected]
ABSTRACT
In today’s scenario considering the covid 19 situation there are lot of security issues the world is facing. In this regard cyber forensic security plays vital role in enhancement of security. Cyber forensic includes physical and behavioural characteristics of the person for identifying an actual user and imposter. Unimodal cyber forensic system provide recognition but contain some limitation such as high error rate, non-universality, noise in sensed data etc. These problems could be overcome by multimodal approaches which combine more than one trait of a user for authentication. Multimodal provides better performance, accuracy, and security over a unimodal cyber forensic system. In this paper a new technique is proposed to improve the security of Cyber forensic system by combining the features of fingerprint system and Iris recognition system.
Keywords: cyber forensics, unimodal, multimodal, fusion
Introduction
Unimodal cyber forensic systems also face major limitations due to noise sensitivity, intra-class variability, data quality, non-universality, and other variables in many real-world applications. In such cases, it does not prove successful in trying to enhance the efficiency of individual matchers.
Multimodal cyber forensic Figure 1.1 aim to mitigate some of these problems by offering several pieces of proof of the same identity. A multimodal cyber forensic method to resolve the limitations is introduced in this paper by using several pieces of proof of the same identity.
Figure 1.1 Overview of multimodal system
However, the multimodal cyber forensic device is restricted to time constraints due to its multiple processing phases. Application of Digital Technologies to Agriculture is an innovative idea focusing on the enhancement of agricultural development in rural areas [3].
A dynamic fingerprint identification technique fused with improved iris recognition sing the adaptive rank level fusion method is implemented to increase the speed of authentication in the cyber forensic system with appropriate accuracy.
Various fusion techniques have been applied, including the highest rank, borda count and logistic regression methods. When checked on the standard biometric dataset, the scheme demonstrates changes in the False Acceptance Rate (FAR) and Equivalent Error Rate (EER) curves. Many individuals trust that biometric frameworks can identify liveness in biometric tests [2].
Fingerprint System
It is essential to build security solutions by adopting a Security Framework for any organization to find solutions for majority of vulnerabilities and flaws [4]. Secure data transmission is the fundamental need for the users of internet community [6]. Depending on which one of the following aspects of the problem is being studied, the question of fingerprint individuality can be formulated in several different ways: I the problem of individuality may be cast as assessing the likelihood that any two individuals in a given target population will have sufficiently similar fingerprints assess the likelihood of finding a sufficiently similar fingerprint in the target population when the input of a sample fingerprint is given. We have performed fingerprint matching in two steps: i) point-wise match and ii) trim false matches with arithmetical constraints [1].
The original image in Figure 1.3(a) shows that the central fingerprint area displays a very high variance value, whereas the regions outside this area display a very low variance. Therefore, a variance threshold approach is used to separate the foreground fingerprint area from the background regions.
As shown in Figure 1.3(c), the final segmented image is generated by assigning a grey-level value of zero to the regions with a variance value below the threshold. When deciding the threshold value used to segment the picture, there is a trade-off involved. Results have shown that foreground regions can be wrongly allocated as background regions if the threshold value is too high.
Conversely, if the threshold value is too small as part of the fingerprint foreground area, background regions may be mistakenly assigned.
Fig.1.2
Therefore, the variance threshold technique is effective in discriminating the foreground area from the background regions. When deciding the threshold value used to segment the picture, there is a trade-off involved. Results have shown that foreground regions, if the threshold value is too large, may be incorrectly assigned as background region. Conversely, if the threshold value is too small, as part of the fingerprint foreground area, background regions may be mistakenly assigned.
A variance threshold of 100 gives an optimal result in terms of distinction between the foreground and background regions.
Fingerprint Image Enhancement
The quality of the ridge structures is an essential feature in a fingerprint picture, as the ridges carry the information of characteristic features needed for the extraction of minutiae. Ideally, the ridges and valleys should alternate and flow in a well-defined fingerprint picture in a constant local direction. Under such a condition, transforming the information from the transmitter to the receiver requires more security.[7].
The identification of ridges is enabled by this regularity and, therefore, minutiae can be precisely extracted from the thinned ridges. In this article, the equalisation of histograms is used for image enhancement.
Histogram equalization
The histogram of an image reflects the relative frequency at which the different grey levels in the image occur. Histogram modelling techniques alter an image in order to have a desired form in its histogram. This is helpful in extending narrow histograms with the low contrast levels of images.
Modelling histograms has been shown to be a effective image enhancement tool.
The security of computer’s stored information is now contingent on the level of security of every other computer to which it is connected. Recent years have seen a growing awareness of the need to improve information security [9]. This mapping extends the contrast (expands the grey level range) near the maximum histogram for grey levels. By extending the contrast for most of the image pixels, the transformation increases the detection potential of many image features.
The quality of the ridge structures in a fingerprint image is an essential feature, as the ridges hold the data of characteristic features needed for minutiae extraction. In a well-defined, preferably, The ridges and valleys should alternate and flow in the constant local direction of the fingerprint picture.
The identification of ridges is enabled by this regularity and, therefore, minutiae can be precisely extracted from the thinned ridges.
Since fingerprints have many conspicuous landmarks, any combination of them could be usd for the establishment of a reference point in order to establish a reference point. In a fingerprint image, the point of maximum curvature of the ridges is known as the reference point of a fingerprint.
A reference point as well as the orientation of each image must be located for aligning two fingerprint images. The core point is the most generally used reference point.
A core point is known as the point at which in the orientation field of a fingerprint image a maximum direction change is detected or the point at which the directional field becomes discontinuous. For key point detection, several methods have been suggested.
In order to compute the orientation image, the least mean square estimation method employed by Hong et al (1999) is used.
Binarization
On binary images, where there are only two levels of interest, most minute extraction algorithms operate: the black pixels that represent ridges, and the white pixels that represent valleys.
Binarization is called the conversion of a grey level image into a binary image. This increases the contrast in a fingerprint picture between the ridges and valleys, and thus allows the extraction of minutiae. The binarized image production from the enhanced image
Figure1.3 Results of performing minutiae extraction on a fingerprint image.
The binarization process involves evaluating each pixel's grey level value in the enhanced image, and if the value reaches the global threshold, the pixel value is set to one binary value; otherwise, it is set to zero. The effect is a binary image that comprises two data levels, the ridges in the foreground and the valleys in the background.
Thinning is the final image enhancement stage usually performed before removing minutiae.
Thinning is the morphological process that successively erodes the foreground pixels away until they are one pixel wide.
Iris Recognition
Figure 1.4 Eye Structure and Iris features
The iris is the coloured, clearly visible ring circling the pupil. It is a muscular structure with intricate details that can be measured, such as striations, pits, and furrows, that controls the amount of light entering the eye. The iris, as in Figure 3.16, is not to be confused with the retina, which forms the
In order to use this technology, one has to look at the camera first. Without permission, iris recognition cannot take place. In order to remove only the iris portion, the image of an eye is first processed by software that locates the inner and outer borders of the iris, and the eyelid contours.
Reflections and eyelashes that can cover It distinguishes and discounts portions of the iris. RPM and TLA processing provide final results which are undoubtedly visible for health practitioner reviews for pre and post or even during surgical activities [10].
Iris segmentation is the first step in iris recognition and determines the image material used for extracting and matching features, which is directly related to the accuracy of recognition. In practical applications, speed is always a bottleneck, and iris segmentation is often discovered to be the most time-consuming module in an iris recognition device. Most failures in iris recognition matching are stated to result from incorrect iris segmentation.
This operator acts as a circle finder that scans the radial derivative 's maximum angular integral over the kmeans clustering algorithm on the iris image's location and intensity function vector.
Finally, by testing the connectivity of the candidate points to the upper eyelid, the detection outcome is refined. The notion is that near the upper eyelid, most eyelashes and shadows appear.
This refinement is important because the pressure of choosing the detection threshold is relieved.
This makes it possible for us not to expend too much time trying to find an ideal threshold, but just a moderately successful and loose one.
Multimodal Cyber Forensic System
A higher recognition rate and the effort to boost the performance of single matchers is often not feasible. In such cases, due to inherent issues, a single recognizer cannot prove to be effective.
These problems can easily be alleviated by using a multi cyber forensic method by presenting several pieces of evidence from the same human subject, thereby achieving greater and more accurate identification.
In this paper, to enhance device performance, the results of fingerprint and iris authentication are combined. The raw fingerprint and iris scores have different distributions, and the sigmoid function is applied from 0 to 1. to normalise these raw scores.
Finally, by fusing these normalised scores using an adaptive rank level fusion process, the multimodal cyber forensic authentication system is proposed. To categorise the unknown consumer into acceptance or rejection, the fused score is used.
Normalization is called converting the raw scores obtained using various modalities to a common domain using a mapping function. In general, normalisation has different techniques, such as z- score, min-max, decimal adjustment, and sigmoid function. In this article, the sigmoid feature is used to normalise fingerprint and iris raw scores. In this paper, the sigmoid function is used as it is beneficial to include the outlier data while also maintaining the value of data within the standard deviation of the mean.
Fusion can be performed at the rank level when a subset of possible matches sorted in decreasing order of trust are the output of each cyber forensic matcher. In order to derive a consensus rank for each identity, the objective of rank level fusion is to consolidate the rank output by individual cyber forensic subsystems (matchers).
This technique of rank fusion is close to applying the max rule at the score stage for fusion.
Relations between users can be randomly broken as a consequence of applying this fusion law.
Conclusion
The different rank level fusion techniques such as highest rank, borda count and logistic regression methods are compared in terms of Genuine Acceptance Rate (GAR) to determine the multimodal results of fingerprint and iris. It is obvious from the findings that when compared to the highest rank and borda count methods, the multimodal authentication scheme with logistic regression fusion techniques has better error rates. In addition, it contrasts the preparation and recognition time of different approaches to rank level fusion. Compared to other approaches, it could be concluded that the higher rank approach demonstrates a 10 percent increase in recognition time.
References
[1] Sahithi, S., Anirudh, A., Swaroop, B., Ruth Ramya, K. Biometric security for cloud data using fingerprint and palm print 2019 International Journal of Innovative Technology and Exploring Engineering863383432
https://www.scopus.com/inward/record.url?eid=2s2.085069451595&partnerID=40&md5=6a410 2e306106fd16731338b23028bd9
[2] Tarannum, A., Rahman, M.D. Multi-modal biometric system using Iris, Face and fingerprint images for high-security application 2019 International Journal of Recent Technology and Engineering 76314320 https://www.scopus.com/inward/record.url?eid=2-
s2.085067962719&partnerID=40&md5=b1b1c2acd0ee967c767d7a35cad52cbc
[3] Puvvada, N., Prasad Babu, M.S.Semantic web based banana expert system 2018 International Journal of Mechanical and Production Engineering Research and Development 833643713 https://www.scopus.com/inward/record.url?eid=2-
s2.085062992172&partnerID=40&md5=3579f6c1ea568fcc5ab66dc77cdd7dd1
428 10.1049/iet-ipr.2018.52886
[6] Sahu, Aditya Kumar; Swain, Gandharba Data hiding using adaptive LSB and PVD technique resisting PDH and RS analysis INTERNATIONAL JOURNAL OF ELECTRONIC SECURITY AND DIGITAL FORENSICS 2019 114458476 10.1504/IJESDF.2019.102567
[7] Biometric-based efficient medical image watermarking in E-healthcare application Aparna, P;
Kishore, PVV IET IMAGE PROCESSING FEB 28 2019 10.1049/iet-ipr.2018.528864
[8] Adaptive PVD Steganography Using Horizontal, Vertical, and Diagonal Edges in Six-Pixel Blocks Pradhan, A; Sekhar, KR; Swain, G SECURITY AND COMMUNICATION
NETWORKS 2017 10.1155/2017/19246184
[9] A Dependency analysis for Information Security and Risk Management Krishna, BC;
Subrahmanyam, K; Kim, THE INTERNATIONAL JOURNAL OF SECURITY AND ITS APPLICATIONS AUG 2015 10.14257/ijsia.2015.9.8.17
[10] Bangare, Sunil L.; Pradeepini, G.; Patil, Shrishailappa: a new computational technique for precise medical imaging INTERNATIONAL JOURNAL OF BIOMEDICAL ENGINEERING AND TECHNOLOGY 2018 27 1-2 76 85
