New Texture Descriptor Based on Modified Fractional Entropy for Digital Image Splicing Forgery Detection.

Area of Science:

• Digital Image Processing
• Computer Vision
• Forensic Science

Background:

• Digital image forgery, particularly splicing, is a growing concern due to advanced manipulation tools.
• Existing algorithms struggle with high-dimension feature vectors for accurate forgery detection.
• Splicing creates anomalies in image features, necessitating robust detection methods.

Purpose of the Study:

• To enhance the accuracy of image splicing detection.
• To develop a method utilizing low-dimension feature vectors.
• To introduce a novel approach for capturing splicing artifacts.

Main Methods:

• Proposed an approximated Machado fractional entropy (AMFE) as a fractional texture descriptor.
• Applied discrete wavelet transform (DWT) to decompose images into frequency sub-bands.
• Evaluated the approach on the CASIA v2 image dataset.

Main Results:

• Achieved superior detection accuracy compared to state-of-the-art methods.
• Demonstrated effective capture of splicing artifacts using AMFE and DWT.
• Obtained improved positive and false positive rates with low-dimension feature vectors.

Conclusions:

• The proposed AMFE-DWT method offers a highly accurate and efficient solution for image splicing detection.
• Low-dimension feature vectors can effectively represent splicing artifacts.
• This technique advances digital image forensics by improving forgery detection capabilities.