A method for analyzing action sensing signals of a doll skeleton

Abstract

The application provides a doll skeleton motion sensing signal analysis method, comprising: calculating the axial touch position according to the radial component amplitude and the tangential component amplitude in the initial touch sensing signal set, and evaluating the filling compression accumulation degree in combination with the real-time flexion angle; by jointly analyzing the modified component amplitude set and the real-time flexion angle, using the k-means clustering algorithm to divide the signal type boundary and identify the signal type drift direction; according to the signal type drift direction, dynamically updating the judgment threshold range of light touch and extrusion to obtain the posture adaptability judgment criterion; evaluating the matching degree of the posture adaptability judgment criterion and the initial touch sensing signal set, fusing the modified component amplitude set to perform waveform comparison, and determining the touch action type; using the determined touch action type as input to generate a consistency discrimination result output, and realizing unified recognition of touch actions under different skeleton postures.

Images