Abstract
Neophyte finger print minutiae are producing extremer forensic evidence for commercial criminal inquiries. Such minutiae are hard to identify from fingerprint images due to extremely small size and irregular shape minuscule. In that method, we pioneered the deep learning based segmentation, AGNN classifier, and wavelet feature for fingerprint triviality detection project. Then, preprocess fingerprint images through contrast enhancement, demising, and image segmentation for isolating ROI where the fingerprint will be present alone, and then create a deep learning model based on Convolutional Neural Networks (CNNs) that will learn the features of those fingerprints and label them as trivial or non-trivial. Apart from that, employ the AGNN model, which employs the learned spatial relationships between the trivialities in the fingerprint, to determine whether the photographs belong to the trivial or non-trivial category. Lastly, wavelet analysis will be employed to obtain features that reflect the texture and form of the trivialities from the fingerprint image. The model first applied image segmentation algorithms to obtain minutiae from fingerprint images. Then, the AGNN classifier, an attention-based graph neural network, will be employed for classification of minute based on their derived features. In addition, these wavelet features are also derived from the fingerprint image to enhance the accuracy of classification. The system proposed anticipates an increase in accuracy and reliability in fingerprint identification systems and has extensive applications in the security sector.