Children biometric recognition for the betterment of society using deep intelligence
V. Kamble, Manisha P. Dale · IET conference proceedings. · 2022
Missing and abduction of children is a big social issue nowadays. According to the National Crime Records Bureau, India reports more than 100 children go missing every day. Children are in the developing stage till their adulthood. Their biometrics also go through many changes. Therefore, biometric identification system for children is not yet developed. Children's recognition using their biometrics is a challenging task. Many researchers are working on this problem using different biometrics, such as finger, foot, palmprint, face, iris, ear and headprint. However, the issue is not yet resolved satisfactorily. The problem in recognizing children is the small size of biometrics. It is under developing stage, therefore displacement of key features observed, which leads to false acceptance. In this research the footprint biometric is used for identification of children. Most of the work done in literature is in verification mode. In this paper biometric identification is performed which will be helpful in finding the missing children. The database of 50 children's footprint is collected. The age of the children is 1 month to 5 years. Two types of input images are compared to achieve high accuracy. First input images are obtained by pre-processing and Region of Interest (ROI) extraction of the crease pattern from footprint images. Second are obtained by texture-based enhancement of crease pattern's ROI. Both the inputs are compared using state-of-the-art techniques and modern deep learning algorithms. The deep learning algorithm with enhanced crease patterns is giving 100 % accuracy, which is higher than the literature.