Twin face recognition using a deep learning-based pixel difference network with edge maps

• Published in: Signal, image and video processing image and video processing, 2025, Vol.19 (1)
• Main Authors: Kavitha, P., Ahilan, A., Gnanamalar, A. Jasmine, Usha, M.
• Format: Article
• Language: English
• Subjects: Classification; Computer Imaging; Computer Science; Computer vision; Deep learning; Edge detection; Face recognition; Facial recognition technology; Feature extraction; Image filters; Image Processing and Computer Vision; Machine learning; Multimedia Information Systems; Original Paper; Pattern recognition; Pattern Recognition and Graphics; Pixels; Signal,Image and Speech Processing; Vision
• ISSN: 1863-1703; 1863-1711
• DOI: 10.1007/s11760-024-03577-4

Twin face recognition system is one of the most challenging areas of pattern recognition and computer vision. There are several challenging aspects in the twin facial recognition technology, such as similar faces, eyes, noses, fingerprints, etc. In this research, a novel deep learning (DL)-based PD-IRNet method has been proposed for classifying twins. Adaptive Unsharp mask Guided filter (AUG) is utilized to pre-processed for reducing the noise artifacts in images. Then, the image scaling is performed to scale the image and adjust the pixel height and width images. Then, the feature extraction process uses a Pixel Difference Network based on Edge Detection to extract the edges. The edge features are extracted and based on these features DL-based Inception–ResNet is used to classify into identical and non-identical twins. The experimental PD-IRNet achieves overall accuracy of 98.45% from the gathered dataset. In comparison with the current methods, the proposed PD-IRNet model achieves an overall accuracy of 98.45%.