iris.nodes.templates_aggregation package

Submodules

iris.nodes.templates_aggregation.majority_vote module

Iris Template Combiner

This module implements a hybrid approach for combining multiple iris templates from the same user to enhance recognition performance.

The algorithm combines: 1. Selective Bits Fusion - Using the most reliable bits 2. Majority Voting with Weight Templates - For consistent bits 3. Inconsistent Bit Analysis - To utilize the pattern of inconsistent bits

Author: Rostyslav Shevchenko Date: May 22, 2025

class iris.nodes.templates_aggregation.majority_vote.MajorityVoteAggregation(consistency_threshold: float = 0.75, mask_threshold: float = 0.01, use_inconsistent_bits: bool = True, inconsistent_bit_threshold: float = 0.4, callbacks: List[Callback] = [])

Bases: Algorithm

Class for combining multiple iris templates from the same user.

The algorithm identifies consistent and inconsistent bits across templates, uses majority voting for consistent bits, creates a weight template to reflect bit reliability, and generates a combined mask representing consensus of valid regions.

class Parameters(*, consistency_threshold: ConstrainedFloatValue = 0.75, mask_threshold: ConstrainedFloatValue = 0.01, use_inconsistent_bits: bool = True, inconsistent_bit_threshold: ConstrainedFloatValue = 0.4)

Bases: Parameters

consistency_threshold: float

inconsistent_bit_threshold: float

mask_threshold: float

use_inconsistent_bits: bool

combine_templates(templates: List[IrisTemplate]) → Tuple[IrisTemplate, ndarray]

Combine multiple iris templates from the same user.

Parameters:

templates (List[IrisTemplate]) – List of IrisTemplate objects from the same user

Returns:

Combined IrisTemplate weights (np.ndarray): Weight matrix reflecting bit reliability

Return type:

combined_template (IrisTemplate)

run(templates: List[IrisTemplate]) → Tuple[IrisTemplate, ndarray]

Combine multiple iris templates from the same user.

Parameters:

templates (List[IrisTemplate]) – List of IrisTemplate objects from the same user

Returns:

Combined IrisTemplate weights (np.ndarray): Weight matrix reflecting bit reliability

Return type:

combined_template (IrisTemplate)

Module contents