A contactless identification system based on hand shape features

Resumen

This paper aims at studying the viability of setting up a contactless identification system based on hand features, with the objective of integrating this functionality as part of different services for smart spaces. The final identification solution will rely on a commercial 3D sensor (i.e. Leap Motion) for palm feature capture. To evaluate the significance of different hand features and the performance of different classification algorithms, 21 users have contributed to build a testing dataset. For each user, the morphology of each of his/her hands is gathered from 52 features, which include bones length and width, palm characteristics and relative distance relationships among fingers, palm center and wrist. In order to get consistent samples and guarantee the best performance for the device, the data collection system includes sweet spot control; this functionality guides the users to place the hand in the best position and orientation with respect to the device. The selected classification strategies - nearest neighbor, supported vector machine, multilayer perceptron, logistic regression and tree algorithms - have been evaluated through available Weka implementations. We have found that relative distances sketching the hand pose are more significant than pure morphological features. On this feature set, the highest correct classified instances (CCI) rate (>96%) is reached through the multilayer perceptron algorithm, although all the evaluated classifiers provide a CCI rate above 90%. Results also show how these algorithms perform when the number of users in the database change and their sensitivity to the number of training samples. Among the considered algorithms, there are different alternatives that are accurate enough for non-critical, immediate response applications.