New Method of Intelligent Control Based on Multi-Information Fusion
This paper advances a new method for intelligent control which based on face and human posture recognition.The algorithm first performs identity authentication through face recognition,then collects human bone information based on depth images for gesture recognition,and finally performs instruction reconstruction based on the recognized gesture information.Among them,the face recognition part adopts the newer LTDP2 algorithm(second-order local ternary derivative pattern).Experiments have verified that the LTDP2 algorithm can efficiently extract facial features,and can still maintain a high recognition accuracy rate in complex light scenes.The posture recognition part adopts the trinocular camera to collect the depth information of the human body,obtains the posture instruction through the analysis of the relative position of the bone points,then form a complete set of control instructions by componentized refactoring the single instruction.This algorithm can effectively solve the problem of controlling intelligent systems in closed unmanned scenarios.
face recognitionLTDP2depth imagesgesture recognitioninstruction componentized refactoringintelligent control
NETL
NSTL
万方数据
