Research And Implementation Of Gesture Recognition Based On Depth Data

Gestures can convey information not only intuitive,vivid,also can be combined into a variety of forms to convey information.It has broad prospects for development in the smart driving,smart home,auto industry and other fields.Especially in the field of smart home,gesture recognition technology can define a more intelligent,more user-friendly way of interaction,it has been a hot research issue.Initially,gesture recognition algorithm was based on color image expansion,but color images are easily affected by external environment such as light,shadow,color and so on.Microsoft released the Kinect depth camera in 2010,which captures both color images and depth images.The depth of the image contains three-dimensional distance information to solve the lighting,shadows and other environmental impact.Therefore,the main research content of this subject is to design a gesture recognition algorithm suitable for smart home with Kinect depth camera,and to research a gesture recognition algorithm with high real-time,high accuracy and good robustness.This article gesture recognition algorithm is mainly divided into two parts: hand-joint point positioning algorithm and gesture classification and recognition algorithm.Hand point positioning algorithm focuses on the hand position recognition.Based on the study of the structure of opponent's hand and the hand movement characteristics,a hand point algorithm based on random forest is proposed.The Kinect camera is used to collect the hand depth images,and the hand division scheme is designed.The depth image features are analyzed.The depth features are extracted according to the hand position.The random forest algorithm is introduced to classify the image pixels,and the hand joint positioning is completed by category classification.Based on the three-dimensional position of the hand joints,the hand skeleton model is established,the structure vector describing the hand movement is designed,extracting angle and distance composition to describe gesture of the structure vector.The movement of the angle and the relative position of the feature vector changes,the introduction of SVM to achieve the classification of different gestures.According to the application scene of this paper,a smart home simulation experiment interactive system is set up and a set of gesture interaction actions designed in a smart home is designed.Experiments show that the proposed algorithm can accurately and quickly classify the image pixels and accurately locate the position of the hand joints.The classification and recognition algorithm has high accuracy and robustness for complex gesture recognition and can meet real-time requirements.