An Offline Handwriting Age Range Prediction System Using an Optimized Deep Learning Technique

One of the distinguishing features of an individual is handwriting and it has been established that everyone has unique handwriting differing from one another. This unique feature evolves with time and is influenced by a variety of factors such as gender, physical and mental health, and age among others. Also, the recent development in using individual peculiar features for forensic investigations in banks and other allied institutions either for signature verification or identification spelled the need to develop a smart system that can predict the age range with offline handwriting recognition. It is on this background that this research employed an optimized deep learning technique comprising of Gravitational Search Algorithm and Convolutional Neural Network (CNN-GSA) for offline handwriting age range prediction. A local database was populated with samples of the signature captured with a digital camera (5 megapixels), the CNN was employed for feature extraction, GSA was utilized to select optimal CNN parameters used for classification while the combined CNNGSA was utilized for an offline handwritten-based age prediction system. The performance evaluation of the approach proposed was done using sensitivity, specificity, precision, false positive rate, recognition accuracy, and processing time for all the variants, while the superiority of the system developed was ascertained by comparing it with the original CNN.