Classification of finger grasping by using PCA based on best matching unit (BMU) approach

Abstract

In this paper we proposed to analyze in depth the thumb, index and middle fingers on the fingertips bending or grasping movement against an objects. The finger movement data are measured using a low cost DataGlove “GloveMAP” which is based on fingers adapted postural movement of the principal component. In supervised classification, we are provided with a collection of grasping feature whereas the features capable to be categorized using the EigenFingers of the fingertips bending or grasping data. The classification of the fingers activities is analyzed using Principal Component Analysis (PCA) for feature extraction or normalization reduction and is used for fingertips movement dataset. Meanwhile for the finger grasping group features, the method of Best Matching Unit (PCA-BMU) was proposed whereas the concept of Euclidean Distance could be justify by the best grouping features according to the best neuron or winning neuron. The use of the first and the second principal components can be shown in the experimental results that allow for distinguishing between three fingers grasping and represent the features for an appropriate manipulation of the object grasping.