A Conformable Moments-Based Deep Learning System for Forged Handwriting Detection.

Area of Science:

• Computer Vision
• Machine Learning
• Pattern Recognition

Background:

• Handwriting forgery detection is crucial for machine learning applications.
• Degraded image quality (noise, blur) presents significant challenges.
• Existing methods struggle with complex alterations and image degradation.

Purpose of the Study:

• To develop a robust model for forged handwriting detection in noisy and blurry conditions.
• To introduce a novel approach combining conformable moments and deep ensemble neural networks.
• To classify various types of handwriting alterations including noise and blur.

Main Methods:

• Utilized conformable moments (CMs) for preserving image details via fractional calculus.
• Developed a deep ensemble neural network (DENN) classifier with stenographic kernels and spatial features.
• Classified images into normal, altered, noisy, blurred, altered-noise, and altered-blurred categories.

Main Results:

• The proposed CMs and DENNs model demonstrated superior performance in detecting forged handwriting.
• Achieved high classification rates across multiple datasets, including newly introduced and standard ones.
• Effectively handled various image degradations like noise and blur.

Conclusions:

• The integrated approach of conformable moments and deep ensemble neural networks offers a powerful solution for forged handwriting detection.
• The model's robustness in noisy and blurry environments makes it suitable for real-world applications.
• This research advances the state-of-the-art in automated document examination and forensic analysis.