SVM Classifier-Based Automatic Machine Learning Forgery Identification

• Kayode Oke

  English

  Author

• Isiaka Oluwole

  English

  Author

Composites of analog images may be produced by photographers; however, this technique takes a lot of
time and requires specialized skills. Modifications to digital images are made easily using the editing
program. This project examines picture composition, often known as splicing, which is one of the most
popular types of photography alteration. For that purpose, a fraud detection technique is used to take
advantage of minute variations in the color of an image's lighting. Images with two or more persons may
be used with the machine learning approach. This idea is accomplished by using data from statistical
(texture and edge) and physics (chromaticity)-based illumination estimators on picture portions of
related images. A machine-learning technique is then given the extracted texture, skin pigmentation, and
edge-based data for automated decision-making. An SVM (Support Vector Machine) meta-fusion
classifier's classification performance.

• pdf