Script-Identification-Using-Signature-Image-Texture-Analysis

Dataset

Signature Dataset in 4 languages

Dataset Source

Dr. SK MD Obaidullah Google Scholars

Feature Extraction Technique

Gray-Level Co-Occurrence Matrix (GLCM) Link

GLCM Details

• Offsets: 16
• Number of GLCM matrices: 16
• Number of features: 7
• Features: Energy, Homogeneity, Correlation, Contrast (All GLCM Properties), Entropy, MeanG, Standard Deviation
• Total no of features: 16 * 7 = 112

Learning Algorithm

• Multilayer Perceptron
• Logistic Regression

Identified Scripts

Bengali, English, Hindi, Urdu

Brief Description

After importing the dataset it was converted into gray scale image from RGB image using in-built matlab function rgb2gray().The image was signature image and already cropped.That’s why in order to extract details no more pre-processing was done.After that glcm was applied and features was extracted.Then the features was delivered to multilayer perceptron algorithm, which gave the final output.

Special Thanks

• Dr. SK MD Obaidullah LinkedIn
• Jipan Ali LinkedIn