CNN-Based Dactylogram Classification for Finger and Palmar Regions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for classifying dactylograms are unsuitable for distinguishing between various types of dactylograms, such as finger and palmar dactylograms, leading to potential errors in classification and integrity issues in databases, especially in multimode acquisition campaigns.
Innovation Solution
A computer-implemented method using a convolutional artificial neural network trained on a dataset of dactylograms classified into specific anatomical regions of the hand, providing probabilities for each class and allowing for automatic and accurate classification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing classification methods are used, then the process is simple, but classification accuracy deteriorates because they cannot distinguish between finger and palmar dactylograms
Solution Approach 1:
The patent replaces traditional mechanical/image-processing-based classification methods with a convolutional neural network (CNN) that uses deep learning to automatically classify dactylograms. The CNN architecture, comprising multiple convolutional layers with filters, achieves superior classification accuracy by learning hierarchical features from dactylogram images, thereby resolving the contradiction between simplicity and accuracy.
Solution Approach 2:
The patent transforms the classification approach by changing the parameter space from traditional image processing parameters to neural network parameters. The CNN models the dactylograms through multiple convolutional layers, each applying different filter parameters to extract increasingly abstract features, enabling accurate distinction between finger and palmar dactylograms.
2Reliability
If manual classification by operator is performed, then device complexity is low, but reliability deteriorates due to operator errors
Solution Approach 1:
The system implements self-service automation where the convolutional neural network independently performs classification without requiring operator intervention. The automated CNN system processes dactylograms through its convolutional layers and outputs classifications reliably, eliminating human error while maintaining high reliability.
Solution Approach 2:
The patent substitutes the mechanical manual classification process with an automated neural network system. The CNN replaces operator judgment with computational algorithms that consistently apply the same classification criteria, thereby improving reliability and eliminating variability introduced by human operators.
3Loss of information
If dactylograms are not classified by type, then processing is faster, but loss of information increases due to inability to distinguish finger from palmar dactylograms
Solution Approach 1:
The patent applies preliminary action by classifying dactylograms into finger and palmar types before subsequent processing steps. The convolutional neural network performs this classification in advance, allowing the system to then process each type appropriately, thereby preventing information loss while maintaining efficient processing through specialized handling of each dactylogram type.
Data Source
AI summary
A computer-implemented method for classifying dactylograms among a plurality of membership classes in which each of the classes corresponds to a specific anatomical region of the palmar face of a hand. The method includes taking, as input data, at least one dactylogram and providing, as output data, a membership class or a list of membership classes to which the dactylogram belongs from among the plurality of membership classes.