12 - Fingerprints

What

Set of lines on each finger.

Acquired by livescan or inkpad.

To compare if two fingerprints are identical:

  • correlation: the two ones are superimposed and the computation of correlation between corresponding pixels is iterated for different alignments
  • ridge features: ridges orientation and local frequency, shape of ridges and texture, which are more reliable and easier to extract, but also less distinctive
  • minutiae: the minutiae are first extracted from the two fingerprints and stored as two sets of points in a two dimensional space, and then the methods search for the alignment between the two sets that maximizes the number of corresponding pairs of minutiae, and based on this measure the similarity between the fingerprints (point pattern matching)

Problems:

  • movement
  • distortion
  • non-linear distortion of the skin
  • variable pressure and skin condition
  • errors in the features extraction

Main Techniques

  • Minutiae-Based Matching: Focuses on minutiae points (ridge endings and bifurcations) and their spatial relationships.
  • Ridge Pattern-Based Matching: Analyzes the overall ridge flow and pattern structures.
  • Singular Points: Identifies core points and delta points in the fingerprint.
  • Correlation-Based Matching: Compares the correlation between the input fingerprint and stored templates.

Feature Extraction

  • Minutiae Extraction: Identifies and records minutiae points and their angles.
  • Ridge Counting: Determines the number of ridges in a specific area.
  • Local Ridge Frequency Estimation: Analyzes ridge frequency to distinguish between different areas of the fingerprint.
  • Gabor Filters: Utilizes Gabor filters for frequency and orientation analysis.

How Feature Extraction Works

  • Minutiae Extraction: Algorithms locate ridge endings and bifurcations, recording their coordinates and angles.
  • Ridge Counting: Counts ridges in predefined blocks, providing local ridge density information.
  • Local Ridge Frequency Estimation: Divides the fingerprint into blocks and analyzes the frequency of ridges in each block.
  • Gabor Filters: Gabor filters are applied to enhance ridge features at specific frequencies and orientations.

Frequency and Directional Maps

  • Frequency Map: Represents the local ridge frequencies across the fingerprint. High frequencies indicate ridge endings, while low frequencies indicate ridge bifurcations.
  • Directional Map: Captures the ridge orientation angles. These maps are essential for orientation estimation and minutiae extraction.