Title：Feature pre-aligned random forest classifier for gesture recognition and classification method
Country：China
Patent No.：201910381675.8
Legal Status：Authorized
Inventor：Xiangyu li, Feifei Zhou
Assignee：Tsinghua University
Address：Tsinghua University,Haidian District Beijing 100084, China
Filing Date：2019-05-08
Issue Date：2021-05-28
Abstract：
The invention discloses a radar-based classifier for gesture recognition. Through signal processing, a range Doppler map (RDM) of each frame of signal is obtained, so that an RDM sequence is obtained,and characteristics are extracted for gesture recognition. In order to solve the problem of feature dislocation caused by distortion of gesture data in time dimension, the invention provides a feature pre-alignment gesture recognition algorithm. The method comprises the following steps: firstly, c on the basis of DTW (Dynamic Time Warping), generating a template for each category, then aligning the features of each frame according to the template of each category, training a binary classification random forest for judging whether a test sample belongs to the category, and finally classifyingthe gesture according to the probability that the test sample belongs to each category. Experiments prove that the gesture recognition accuracy of the classifier for users who do not provide trainingsamples is improved, the recognition rate for eight different gestures is 91.9%, and meanwhile few training samples and low calculation complexity are needed.
Patent Certificate：
PDF/Jpg