Finger Print Recognition

Description:

The biometric identification system described utilizes the Scale-Invariant Feature Transform (SIFT) and Brute-Force Matcher (BFMatcher) algorithms for feature extraction and matching of fingerprints based on image data.

Modality:

The system operates based on image data, processing images from both a test dataset and a training dataset. Each image contains unique features that are extracted and compared to identify matches between test and training images.

Algorithms Used:

• Scale-Invariant Feature Transform (SIFT):

  SIFT is employed for extracting distinctive key points and generating descriptors from images. These descriptors are robust against scale and rotation variations, making them suitable for accurate feature matching.

• Brute-Force Matcher (BFMatcher):

  BFMatcher is utilized for performing brute-force matching between the descriptors extracted from test and training images. It compares each descriptor from the test image against every descriptor from the training image to find matches.

System Workflow

Data Input:

• The system takes image data from two datasets: a test dataset (containing images to be identified) and a training dataset (containing reference images).

Feature Extraction:

• SIFT algorithm is applied independently to both the test and training images to extract keypoints and compute descriptors that represent their unique features.

Matching Process:

• BFMatcher compares the descriptors obtained from the test image with those from the training images. It performs exhaustive matching to identify corresponding features between images.

Evaluation:

• The system evaluates False Match Rate (FMR) and False Non-Match Rate (FNMR) at various thresholds.

• It constructs a Receiver Operating Characteristic (ROC) curve to visualize the performance across different thresholds.

• The Equal Error Rate (EER) is calculated from the ROC curve to assess the system's accuracy.

Output:

• 

•