Palmprint Identification Using Non-Imaging Database

Numerous applications, from forensic to mobile phones, widely use palmprint recognition technology. In this study, we looked at how much work has been done on imaging databases, where image spoofing is becoming more prevalent every day. To develop an anti-spoofing system, we chose to work on non-imaging databases of palm prints, therefore we employed non-imaging databases in this proposed study, and we acquired 450 samples of palm prints on our own in a multispectral laboratory. Using the view spec 4 program, we performed some preliminary processing on the database, taking the 2nd order derivative and its mean before dividing the data and training the model. Utilize various classification techniques. First, we used KNN to classify data, and the result was 70.65%. Next, we used a machine learning algorithm to classify data using the random forest algorithm. Finally, we calculated the accuracy score using the confusion matrix, the F1 score, precision, and recall, and the result was an overall accuracy of 84%.