Research Of Feature Gatherin Technology For3D Handwriting Recognition

Compared with the traditional2D plane handwriting recognition,3D spacial handwriting recognition is currently a new handwriting recognition technology. It can provide users a more natural and efficient human-computer interaction experience, in recent years gradually becomes the hot spot research of handwriting recognition technology, and the3D spacial handwriting recognition is the future tendency of handwriting recognition technology. At the same time, based on the3D acceleration sensor and because of its small volume, high accuracy, resisting external environment interference ability etc., the study of space handwriting recognition system is more and more be taken seriously in recent years.For3D handwriting recognition, the feature extraction and feature dimension reduction are always the two important steps. Feature fusion algorithm is relatively fixed, but the feature extraction and feature dimension reduction algorithm are agile and changeable, so use different algorithms will produce different results. Therefore the3D handwriting recognition needs to in depth study the related feature extraction technology and feature dimension reduction techniques. Although3D handwriting recognition research has made progress these years, but still needs to be further in these two aspects, this paper puts forward a new method of the two key technologies.This paper summarizes and introduces3D handwritten recognition main feature extraction algorithms and feature dimension reduction algorithms, then a classification identification method based on PCA(Principal Component Analysis) and LDA(Linear Discriminant Analysis) combination is proposed:firstly extract time domain feature--RF(Rotation Feature) feature and the frequency domain feature--FFT(Fast Fourier Transform) feature, then fuse the time domain feature and frequency domain feature, we get RF+FFT feature; secondly using PCA+LDA combination as dimension reduction; finally using SVM (Supported Vector Machine) for classification and recognition. The results show that, using the feature extraction technology and feature dimension reduction technology, we can establish better3D handwriting recognition system, so we can improve the recognition rate of3D handwriting recognition system efficiently.