Hand gesture recognition for human computer interaction, being anatural way of human computer interaction, is an area of active research incomputer vision and machine learning. This is an area with many different possibleapplications, giving users a simpler and more natural way to communicatewith robots/systems interfaces, without the need for extra devices. So, the primarygoal of gesture recognition research is to create systems, which can identifyspecific human gestures and use them to convey information or for devicecontrol. For that, vision-based hand gesture interfaces require fast and extremelyrobust hand detection, and gesture recognition in real time. In this studywe try to identify hand features that, isolated, respond better in various situationsin human-computer interaction. The extracted features are used to train aset of classifiers with the help of RapidMiner in order to find the best learner. Adataset with our own gesture vocabulary consisted of 10 gestures, recordedfrom 20 users was created for later processing. Experimental results show thatthe radial signature and the centroid distance are the features that when usedseparately obtain better results, with an accuracy of 91% and 90,1% respectivelyobtained with a Neural Network classifier. These to methods have alsothe advantage of being simple in terms of computational complexity, whichmake them good candidates for real-time hand gesture recognition.
展开▼
