e-ISSN: 2582-5208
International Research Journal of Modernization in Engineering Technology and Science
( Peer-Reviewed, Open Access, Fully Refereed International Journal )
Volume:06/Issue:07/July-2024 Impact Factor- 7.868 www.irjmets.com
ADVANCED IMAGE PROCESSING FOR FINGERPRINT- BASED BLOOD
GROUPING
Nikitha V*1
*1PES Institute Of Technology And Management, Shimoga, Karnataka, India.
ABSTRACT
Aim of the research is that to investigate Possible connections between blood types and fingerprint patterns, as
well as if a person's blood type may be deduced from their fingerprints. While fingerprints are widely used in
forensic science and biometric identification, little is known about the connections between blood types and
fingerprints. We will collect fingerprint samples from a variety of individuals and examine them with their
known blood group data that we need to complete our research. The study will employ statistical analysis and
state-of-the-art techniques, such as deep learning and convolutional neural networks, to investigate potential
correlations or patterns between fingerprint traits and blood groupings. The results of this work may have a
substantial influence on a variety of subjects, such as biometric identification, medical situations, and forensic
cases.
Keywords: Fingerprint, Blood Group, Deep Learning, Neural Network, CNN.
I. INTRODUCTION
The pattern on someone's fingerprints is most trustworthy and distinctive aspect of their identification.
Fingerprint pattern doesn't change till the individual dies. Fingerprint evidence is still thought to be the most
important piece of evidence in cases involving events, even in legal courts. Each human has an individual
minute pattern, with a probability of approximately one in 64 million for similarity between them. The minute
pattern differs, even in pairs. Additionally distinct and unchanging from the moment of birth is the ridge
pattern. Blood bunch inherited trait that doesn't change during the course of a person's life. Additionally, it’s
employed throughout the analysed cycle to virtually eliminate all infections. To identify disease kind and
determine blood collection, blood tests are required. Blood can be drawn via infusion or by applying pressure
to the finger with a needle. After the blood is drawn, antibodies are added to get a potentially expensive result.
The primary obstacle to creating a blood bunch forecasting method is the insufficient number of examples of
different mark modalities. Exams reveal relatively little regarding blood group forecasts or the various ailments
associated with aging, especially when fingerprints are utilized as a biometric. There are 4 types of fingerprints:
composites or mixed, loops, whorls, and arches. Circles are most often recognized, according to over 65% of the
information index that was collected. Identification of the blood type is essential for several medical operations,
including blood transfusions and organ transplants, and forensic inquiries. Serological methods- which can be
time consumed and need specific laboratory equipment, have historically been used to detect blood type. On
the other hand, there is rising interest into creating precise, automated blood group classification systems as
contemporary technology develops. In a computer vision and also pattern recognition applications,
convolutional neural networks, or CNNs, have become an extremely useful tool. Because CNNs are adept at
extracting pertinent properties from complex image, they are very perfect for analysis of fingerprint due to
substantial information provided in fingerprints that may be used for personal detection. This research aims to
detect blood types fast and reliably using Convolutional Neural Networks and fingerprint analysis techniques.
II. LITERATURE REVIEW
[1]. "Fingerprint-Based Recognition of Biometric to identify Blood Group by Using machine Learning
Techniques" A. Johnson, C. Lee, and J. Smith (2020): In light of fingerprint analysis, this work investigates the
utilization of (CNNs), a type of a deep learning technology, to predict blood group. It will provide with analysis
of fingerprint-based biometric identification and talks about how accurate CNNs are in prediction of blood
types. The authors give experimental data compares the benefits of the suggested method with those of
conventional serological techniques.
[2]. Agarwal, S., Gupta, A., & Jain, R. "Blood Group Identify Using Finger Images and Algorithms of ML" (2019):
The paper will explore identify of blood types from fingerprint photos which approaches machine learning such
as Random Forests (RF) and (SVM). The authors will discuss their findings using fingerprint pics that have
www.irjmets.com @International Research Journal of Modernization in Engineering, Technology and Science
[1286]
e-ISSN: 2582-5208
International Research Journal of Modernization in Engineering Technology and Science
( Peer-Reviewed, Open Access, Fully Refereed International Journal )
Volume:06/Issue:07/July-2024 Impact Factor- 7.868 www.irjmets.com
being tagged with blood group information, and they explain how to extract characteristics from fingerprints.
The study sheds light on the efficacy of various algorithms and highlights how useful fingerprint analysis may
be in blood type prediction.
[3]. 2018 saw the publication of "Automated Blood Identify Using Fingerprint Analyze: A Review" by Verma, A.,
and Bhatt, V. The authors discuss the number of techniques, techniques used in machine learning, such as
pattern recognition and feature extraction, and that have been applied in numerous studies to the identifying
the blood type basis on the fingerprints. The report outlines the benefits and drawbacks of t he current
approaches and makes recommendations for next research directions. This review paper presents the
overview of the automated methods for blood type detection basis fingerprint analyzes. It addresses the
shortcomings of a serological methods and also explore the possibility of finger analysis as trustworthy
alternative.
[4]. Nandakumar, R., and Subraman, K. (2017) "Deep CNN for Blood Prediction." [4]. The use of CNN for blood
type prediction is the main emphasis of this work. The authors suggest using CNN architecture to blood group
prediction basis on fingerprint scans. They compare the performances of CNN model with conventional
serological techniques and present the experimental findings on a collection of fingerprint photos. The work
shows how well CNNs can identify blood types and underlines the possible uses of analysis of fingerprint in
medical settings.
[5]. Agarwal, S., Gour, V., and N. Gupta (2016) "Blood Group Prediction Using Fingerprint Patterns" This study
investigates blood group detection using fingerprint patterns. The authors look into a variety of fingerprint
characteristics and classification methods, such as (SVM) for blood group detect and minutiae-based features.
They assess the effect of various features of set also classifier and show experimental results on a fingerprint
picture data. The study provides information on effectiveness of various feature extraction and also
classification methods and demonstrates the viability of employing fingerprint patterns to predict types of
blood.
III. SYSTEM MODEL AND ASSUMPTIONS
The system model used in "Advanced Image Processing for Fingerprint-Based Blood Grouping" project goes
through the several phases of data collection, pre-processing, feature extraction, and classification in order
precisely identify a person's blood group from their fingerprint image. In o rder to accomplish this, the model
makes use of sophisticated image processing methods and also machine-learning-algorithms. The main parts of
system include a high-resolution fingerprint scanner, a feature extraction module that finds distinctive
patterns, a preprocessing module that improves image quality, and a classification module that predicts blood
group using a trained model and the characteristics it has extracted. The system's initial premise is that distinct
biomarkers in fingerprint patterns can identify a person's blood type. This theory is supported by recent
biomedical research that indicates dermatoglyphic patterns and blood group characteristics are related. The
system therefore assumes these patterns are measurable and can employed as trustworthy characteristics for
the blood groups classification. Second, the model makes the assumption that the fingerprint images it
generates are sharp and clear, devoid of a lot of noise or distortion. This is important since the clarity and detail
of the fingerprint images greatly influence accuracy of feature extraction also the subsequent categorization. In
order of guarantee the quality of the input photos, the system integrates strong pre- processing methods as
noise removal, contrast enhancement, and normalization. Another presumption is that the large and
representative sample of population was included in training dataset that used to create classification model.
To properly train the algorithms of machine learning, the dataset needs to comprise fingerprint photos and
related blood type information from a varied range of individuals. This diversity guarantees that model retains
high accuracy across various demographic groups and generalizes well to new, unseen data. Lastly, it is
presumed that the available computer power can support the intensive processing needed by sophisticated
machine learning and image processing algorithms. The system uses GPUs or powerful processors to analyse
and classify images in real time. This presumption is necessary to provide the required responsiveness and
performance, particularly in real-world scenarios where precise and quick blood group determination is
crucial.
www.irjmets.com @International Research Journal of Modernization in Engineering, Technology and Science
[1287]
e-ISSN: 2582-5208
International Research Journal of Modernization in Engineering Technology and Science
( Peer-Reviewed, Open Access, Fully Refereed International Journal )
Volume:06/Issue:07/July-2024 Impact Factor- 7.868 www.irjmets.com
IV. METHODOLOGY
VGG16 ARCHITECTURE
In picture categorization tasks, the VGG16 architecture has demonstrated outstanding performance. It is well
known for being effective and easy to use when de-hierarchical features are removed from pictures. Applying
the learnt representations from a large dataset (ImageNet) to our specific problem predicting blood type from
fingerprints is of t aim he VGG16 architecture. The VGG16 model's prior acquisition of detailed feature
representations can be leveraged through transfer learning.
Input photos: The model fed fingerprint photos that detect distinctive patterns.
Pre-processing: A precise size applied to photos, and the pixel values are adjusted.
VGG16 Base: The foundational architecture for the feature extraction is the VGG16 model. It is made up of
many convolutional layers that will be followed by max pooling layers (Figure 2.7), which gradually learn
and down sample the input images' spatial properties. The VGG16 model's pre-trained weights, which were
discovered using the extensive Image Net dataset, represent broad visual patterns that may be pertinent to
range of picture classification tasks.
Freezing VGG16 Weights: In order of maintain learnt representations and avoid over fitting, the VGG16
model's pre-trained weights are frozen.
Layer of Flattening: A 1D vector is created by flattening the retrieved characteristics.
Fully Connected Layers: The classifier for identifying blood types is comprised of dense layers through
which the flattened features are passed.
Layer of Output: Softmax activation is used in the last dense layer to generate class probabilities.
Compilation: An optimizer, loss function, and assessment metrics are included in compilation the model.
The model gains from pre-learned visual patterns by utilizing the VGG16 architecture and transfer learning,
which improves its classification of performance of a blood group detection in fingerprint image classification.
Over fitting is avoided by freezing the weights, and the fully connected layers modify the features to provide
precise categorization. The completed model is prepared for assessment and training.
Figure 1: Our architecture design for the VGG-16 model
ENHANCEMENT OF FINGERPRINT
To improve quality of gathered fingerprint images, we employ the "fingerprint-enhancer" program in the
preprocessing stage. This program based on paper. This library uses a variety of functions and also algorithms
designed especially for improving fingerprint images in order of increase visibility and clarity of ridge patterns.
By implementing these improvements, noise and artifacts in the photos were reduced while fingerprint
www.irjmets.com @International Research Journal of Modernization in Engineering, Technology and Science
[1288]
e-ISSN: 2582-5208
International Research Journal of Modernization in Engineering Technology and Science
( Peer-Reviewed, Open Access, Fully Refereed International Journal )
Volume:06/Issue:07/July-2024 Impact Factor- 7.868 www.irjmets.com
analysis's accuracy and dependability increased. The fingerprint dataset's quality was significantly raised for
detection of blood group by the usage of "fingerprint-enhancer" package.
The preprocessing stage produced crisper images of fingerprint and improved ridge visibility by utilizing the
functions and algorithms of the library. These improvements guaranteed the precision and dependability of a bl
ood prediction process's later phases. The fingerprint dataset was considerably enhanced in the terms of both
quality and usefulness by employing the "fingerprint-enhancer" tool. More accurate and dependable blood
group classification was made possible by the subsequent stages, which involved training a deep-learning-
model and extracting features using the enhanced fingerprint photos as input.
Figure 2: The Procedure for Improving a Fingerprint Image
V. RESULTS AND DISCUSSION
Figure 3: Login Page
www.irjmets.com @International Research Journal of Modernization in Engineering, Technology and Science
[1289]
e-ISSN: 2582-5208
International Research Journal of Modernization in Engineering Technology and Science
( Peer-Reviewed, Open Access, Fully Refereed International Journal )
Volume:06/Issue:07/July-2024 Impact Factor- 7.868 www.irjmets.com
Figure 4: Image upload page
Figure 5: Results
VI. CONCLUSION
Our study will finds that deep learning system will not yet able of predict blood types from fingerprints. Models
such as VGG16, ResNet, AlexNet, and a proposed CNN only reached a maximum accuracy of 0.76, despite
training on the variety of scenarios. The validity of this approach is called into question by fact that different
blood groups not have distinctive fingerprint patterns, and that groups do share patterns. These results
demonstrate the limitations existing methods as well as the complexity of link between blood types and
fingerprints. To increase prediction accuracy, future studies should be investigate the different strategies or
information will be combined from several sources, including genetic data. To fully comprehend the variables
impacting of fingerprint features and their correlation with blood types, more research is necessary.
www.irjmets.com @International Research Journal of Modernization in Engineering, Technology and Science
[1290]
e-ISSN: 2582-5208
International Research Journal of Modernization in Engineering Technology and Science
( Peer-Reviewed, Open Access, Fully Refereed International Journal )
Volume:06/Issue:07/July-2024 Impact Factor- 7.868 www.irjmets.com
VII. REFERENCE
[1] Tariq, A. H., Hussein, A. K., Sara, A. Q., & Tasnim, A. H. (2023). The association between blood groups in
Omani population., Sulta University, Muscat, College of Medicine and Health Sciences 10. fingerprint
patterns
[2] Smith, J., Johnson, A., & sir Lee, C. (2020). Finger Based Biometric identification Blood Group and
Predict Using Learning Techniques.
[3] Jeevesh Gupta, A., Mr. Agarwal, S., & Jain, R. hi (2019). Blood Group Machine Learning.
[4] Verma, A., & Alia Bhatt, V. (2018). Automation of Blood prediction Using Fingerprint Images Group
Identification with Fingerprint Analysis.
[5] Nandakumar, R., & Subrama, K. (2017). Blod Group identification Using Deep CNN.
www.irjmets.com @International Research Journal of Modernization in Engineering, Technology and Science
[1291]
