Real-time hand gestures system based on leap motion

摘要

In the three-dimensional human-computer interaction, the identification of dynamic and staticgestures is a very important and challenging work in the field of machine vision, In this paper, wepropose a new gesture recognition system. Leap Motion device is a kind of equipment, which isspecially used for hand recognition, which can get the feature data to realize the gesture recognitionin real time. The system ismainly composed of the following two parts. For static gestures,we use a kind of feature information based on the distance, direction, and bending degree ofthe fingertip, and bring the support vector machine into the training to realize the static gesturerecognition. For dynamic gestures, we use gesture length as a benchmark to reject non-key gesturesand preprocess frames with abnormal gesture sequences. The average recognition rate ofstatic gestures reaches 99.98%, andthe recognition rateofdynamicgestures reaches 96.20%.Theexperimental results show that the algorithm has a good effect on gesture recognition, and it issuitable for the simple interaction between gestures, people and people and daily communicationof daily communication barriers.