ppt for biometric blood group detection A Machine Learning Approach

ProjectWorkPhase2(19CS702) – Review 2
DEPARTMENT OF COMPUTER SCIENCE AND ENGINEERING
SAVEETHA ENGINEERING COLLEGE
(Autonomous Institution – UGC, Govt. of India)
(Affiliated to Anna University, Approved by AICTE - Accredited by NBA & NAAC – ‘A’ Grade - ISO 9001:2015 Certified)
Saveetha Nagar, Thandalam, Chennai-602 105, TamilNadu, INDIA.
BIOMETRIC BLOOD GROUP DETECTION: A MACHINE
LEARNING APPROACH USING FINGERPRINT PATTERNS
Submitted by:
SUVVARI LAKSHMANARAO (212221040168)
SUDHINDEV S (212221040166)
MAHARA JOTHISH E (212221040094)
2021-2025 Batch
TEAM NO: 106
Under the guidance of:
Mr.P.MUTHUVEL,M.E,Ph.D,
Associate Professor,Department of ECE

Agenda
1. Introduction
2. Statement of the Problem
3. Scope of the project
4. Methodology
a. Architectural Diagram
b. Flow
c. Algorithm used
d. Design - Use Case Diagram, Class Diagram
e. Implementation
f. Important code segments
5. Output, Results
6. Conclusions
7. Future work
8. References

Introduction
• Blood group detection is vital for medical diagnostics, transfusions, forensics,
and organ transplants.
• Traditional serological methods require blood samples, trained personnel, and
lab facilities.
• These methods are invasive, time-consuming, and impractical in emergencies
or remote areas.
• Essential for medical diagnostics, blood transfusions, forensic investigations,
and organ transplants.
• Mismatched transfusions can cause severe complications, including hemolytic
reactions and organ failure.
• The proposed system follows a structured pipeline, beginning with fingerprint
image acquisition, followed by pre-processing, feature extraction, machine
learning classification, and confidence scoring

Statement of the Problem
1. Need for Accurate Blood Group Identification
 Essential for blood transfusions, organ transplants, forensic investigations, and
medical diagnostics.
2. Limitations of Traditional Testing Methods
 Relies on serological techniques requiring blood sample collection.
 Requires laboratory facilities and trained personnel.
 Invasive, time-consuming, and impractical in emergencies and remote areas.
3. Challenges in Critical Scenarios
 Emergency medical situations require rapid blood group detection.
 Difficulties in large-scale screening, such as military recruitments and blood
donation camps.
4. Proposed Solution
 Develop a non-invasive blood group detection system using fingerprint
patterns and machine learning.
 Provide an efficient, automated, and accessible alternative to traditional
testing methods.

Scope of the project
• Develop a non-invasive blood group detection system using
fingerprint patterns and machine learning techniques.
• Establish a correlation between fingerprint ridge characteristics
and blood group types.
• Utilize image processing techniques for fingerprint analysis.
• Apply machine learning models, specifically logistic regression,
for classification.
• Train and validate the system using a labeled dataset of
fingerprint images with corresponding blood groups.

Methodology
•Fingerprint Scanning : The user provides a fingerprint image using a
biometric scanner or high-resolution camera.
•Preprocessing : Image quality enhancement techniques such as grayscale
conversion, noise removal, and contrast adjustment are applied.
•Feature Extraction : The system extracts key biometric features(ridge count,
minutiae points, pattern type) are identified.
•Model Training : The system is trained on a dataset containing fingerprint
images labeled with known blood groups. A Logistic Regression model is used to
establish a mathematical relationship between fingerprint features and blood
group types.
•Blood Group Prediction : When a new fingerprint is provided, the model
classifies it into one of the four blood groups (A, B, AB, O) along with the Rh
factor (Positive/Negative).
•Result Display : The detected blood group is displayed on the user interface,
along with the probability score.

Methodology/Architecture Diagram/Flow

Design-Use Case Diagram
• UML DIAGRAMS
• Use case Diagram:A use case diagram is a graph of
actors set of use cases enclosed by a system boundary,
communication associations between the actors and
users and generalization among use cases.

Design-Class Diagram
• Class Diagram:A class is a representation of an object
and, in many ways; it is simply a template from which
objects are created. Classes form the main building
blocks of an object-oriented application.

Algorithms used
The project uses following Algorithms:
– K-Nearest Neighbors (KNN) : it is a simple, intuitive method
used for both classification and regression tasks.
– It is used in supervised learning tasks, where the algorithm
learns from labeled data.
– KNN is a type of instance-based learning algorithm. It doesn’t
explicitly learn a model but stores the training instances.
– KNN's performance degrades with an increase in the number
of features or dimensions in the dataset, as the distance
metric becomes less meaningful in high-dimensional spaces.

Algorithms used
The project uses following Algorithms:
Support Vector Machine (SVM) Model:
– SVM is a supervised learning algorithm used for classification
and, in some cases, regression tasks. It works with labeled data,
where the goal is to find a decision boundary (or hyperplane)
that separates the data into classes.
– SVM is less prone to overfitting compared to other algorithms,
especially when the data has many features, because it
maximizes the margin.
– SVM can also handle non-linearly separable data using the
kernel trick.
– SVM is inherently a binary classifier (it works for two classes),
but techniques like One-vs-One and One-vs-All can be applied
to extend SVM to multi-class classification.

Hardware and software selection
Software Requirements:
Operating System: Windows/Linux/macOS
Programming Language:
Development Tools: Jupyter Notebook / PyCharm / VS Code
Machine Learning Frameworks: Scikit-learn for Linear Regression Model
Hardware requirements:
Processor : Intel i3 or higher
RAM : 8GB or more (recommended for large datasets)
Storage : At least 20GB of free disk space

Implementation
• A dataset of fingerprint images labeled with corresponding blood groups is
collected.
• Extract key features (ridge count, minutiae points, pattern
classification).Apply feature selection for the most relevant features.
• Split data (80% training, 20% testing).Train the logistic regression model with
fingerprint features.
• Input fingerprint, preprocess, and extract features.Apply trained model to
classify into one of four blood groups (A, B, AB, O).
• Assign confidence scores.If below threshold (e.g., 70%), suggest rescan or
manual testing.
• Display blood group and confidence score.Option to save/download/print
result for records.

Important Code segments
from scipy.ndimage import label
from skimage.feature import local_binary_pattern
from skimage.morphology import thin, disk
from scipy.ndimage import convolve
# Replace with the path to your saved model
model_path = r'C:UsersNameDownloadsBlood_group_detection_with_fingerprint_CVIP
mainBlood_group_detection_with_fingerprint_CVIP-mainlogistic_regression_model.pkl'
model = joblib.load(model_path)
return model
Preprocessing steps
def crop_borders(image, top=10, bottom=10, left=10, right=10):
return image[top:image.shape[0]-bottom, left:image.shape[1]-right]

Local Binary Patterns (LBP)
def extract_lbp(image):
lbp = local_binary_pattern(image, P=8, R=1, method='uniform')
lbp_hist, _ = np.histogram(lbp.ravel(), bins=np.arange(0, 10), range=(0, 9))
lbp_hist = lbp_hist / lbp_hist.sum() # Normalize
return lbp_hist.tolist()
Orientation
def extract_orientation(image):
gradient_x = cv2.Sobel(image, cv2.CV_64F, 1, 0, ksize=5)
gradient_y = cv2.Sobel(image, cv2.CV_64F, 0, 1, ksize=5)
orientation = np.arctan2(gradient_y, gradient_x) * 180 / np.pi
orientation_hist = np.histogram(orientation.ravel(), bins=8, range=(-180, 180))[0]

Wavelet Transform Features
def extract_wavelet_features(image):
coeffs = pywt.wavedec2(image, 'haar', level=2)
features = []
for coeff in coeffs:
if isinstance(coeff, tuple):
for subband in coeff:
features.append(np.mean(subband))
features.append(np.var(subband))
else:
features.append(np.mean(coeff))
features.append(np.var(coeff))
return features
def extract_features(image):
features = {}
# Ridge endings and bifurcations
ridge_endings_count, bifurcations_count = extract_ridge_endings_bifurcations(image)

Output
• OUTPUT:
• After browsing for our website, the following page will be shown first.

Output
• OUTPUT:
• Browse and upload an image of fingerprint which should be
predicted.
Home Page

Output
• OUTPUT:
• After predicting the blood group will be displayed.
View Review Page

Conclusion
• The AI-powered Blood Group Prediction System presents an innovative, non-
invasive approach to determining blood groups using fingerprint patterns. By
leveraging deep learning techniques, particularly Convolutional Neural
Networks (CNNs), the system efficiently extracts and classifies fingerprint
features, ensuring high accuracy and reliability.
• The interactive user interface enhances usability, making it accessible for both
medical professionals and the general public. With real-time monitoring and
model optimization, this solution has the potential to revolutionize traditional
blood group detection methods, reducing dependency on conventional blood
tests while improving efficiency in emergency and clinical scenarios.

Future Work
• Enhancing the system by linking it with hospital and blood bank databases for
seamless blood group verification and donor-recipient matching.
• Combining fingerprint analysis with other biometric modalities, such as iris
recognition and palm vein patterns, to further enhance prediction accuracy and
robustness.

References
• M. F. Mahmood, H. F. Jameel, S. M. Yaseen and S. L. Mohammed, "Determination of Blood
Groups Based on GUI of Matlab," 2023 Fifth International Conference on Electrical,
Computer and Communication Technologies (ICECCT), Erode, India, 2023, pp. 1-8, doi:
10.1109/ICECCT56650.2023.10179690.
• T. Gupta, "Artificial Intelligence and Image Processing Techniques for Blood Group
Prediction," 2024 IEEE International Conference on Computing, Power and Communication
Technologies (IC2PCT), Greater Noida, India, 2024, pp. 1022-1028, doi:
10.1109/IC2PCT60090.2024.10486628.
• C. Sivamurugan, B. Perumal, Y. Siddarthha, V. K. Murthi, Y. N. Sankar and Y. Varun,
"Enhanced Blood Group Prediction with Fingerprint Images using Deep Learning," 2024 3rd
International Conference on Automation, Computing and Renewable Systems (ICACRS),
Pudukkottai, India, 2024, pp. 1091-1097, doi: 10.1109/ICACRS62842.2024.10841725.

ppt for biometric blood group detection A Machine Learning Approach

ppt for biometric blood group detection A Machine Learning Approach

More Related Content

Similar to ppt for biometric blood group detection A Machine Learning Approach (20)

Recently uploaded (20)

ppt for biometric blood group detection A Machine Learning Approach