Algorithm based on normal coordinate vectors with 16 segments for the data fusion from hand-written Arabic text...

Area of Science:

• Computer Science
• Artificial Intelligence
• Pattern Recognition

Background:

• Hand-written text recognition is crucial for fields like healthcare and law enforcement where paper notes are common.
• Challenges include diverse writing styles and character deformations, hindering accurate data interpretation and fusion.

Purpose of the Study:

• To develop a novel handwriting recognition approach using coordinate vectors to address deformation and variation issues.
• To enhance the accuracy of automatic handwriting interpretation for diverse applications.

Main Methods:

• A novel approach applying coordinate vectors to identify similarities in deformed handwriting.
• Implementation using 16 segments for detailed character analysis.
• Utilized a machine learning approach with MATLAB, evaluating 22 technique combinations.

Main Results:

• Achieved a high accuracy rate of 92.8% for handwriting recognition.
• Demonstrated effectiveness with ensemble and bagged tree machine learning models.
• The coordinate vector method successfully identified similarities across various deformations.

Conclusions:

• The proposed coordinate vector-based handwriting recognition method shows significant promise.
• This technique offers a robust solution for interpreting handwritten records, particularly in professional settings.
• The high accuracy suggests potential for widespread adoption in data-driven fields.