Designing face resemblance technique using near set theory under varying facial features

Area of Science:

• Computer Science
• Artificial Intelligence
• Pattern Recognition

Background:

• Near Set Theory provides a framework for object similarity measurement based on descriptive features.
• Human perception of object similarity is a key inspiration for Near Set Theory.
• Existing face recognition methods struggle with variations caused by facial expressions and plastic surgery.

Purpose of the Study:

• To propose a novel face recognition approach using Near Set Theory.
• To address challenges in face recognition due to facial expressions and plastic surgery.
• To leverage Near Set Theory for both feature selection and recognition.

Main Methods:

• Utilized Near Set Theory as a feature selector with tolerance classes to identify key facial features.
• Employed Near Set Theory as a recognizer, using prominent intrinsic facial features extracted by a deep learning model.
• Conducted experiments on diverse facial datasets including YALE, PSD, and ASPS.

Main Results:

• Achieved 93% accuracy on the YALE face dataset.
• Demonstrated 98% accuracy on the PSD dataset.
• Obtained 98% accuracy on the ASPS dataset.
• Outperformed several state-of-the-art classification approaches in face resemblance classification.

Conclusions:

• The proposed Near Set Theory-based face recognition method effectively handles variations in facial features.
• Near Set Theory proves to be a powerful tool for both feature selection and classification in face recognition.
• The approach demonstrates superior performance compared to existing state-of-the-art algorithms.