Research On Human Gait Skeleton Model Based On The Fusion Between Vision And Tactility

In some high-traffic public places,customs,banks and airports,effective identification of human bodies and the elimination of terrorist criminals are essential for the safety of citizens.As a long-distance identification method,gait recognition has the advantages of non-intrusive,difficult to hide,and not easy to camouflage,which has attracted wide attention of researchers and has high application value in human behavior research and intelligent monitoring.The model-based gait recognition method is one of the important directions of gait recognition,and has great advantages in fast recognition.Finding the right method to quickly extract and build a high-precision model is the premise and key of such a method.The accuracy,rapidity,manageability and high representation of the model-related data play a decisive role in the subsequent feature extraction and recognition process.Therefore,this paper studies the skeletal model extraction problem in gait recognition,fully integrates the visual data with long-distance,non-contact,high information characteristics and the highly accurate haptic data,and obtains the high-precision bone length model and the skeletal quality model of mechanical features,which provides data support and theoretical basis for gait recognition and prediction.Aiming at the inconsistency between visual measurement and tactile measurement in time and space,a time fusion algorithm based on motion cycle and a spatial fusion algorithm based on spatial transformation are proposed respectively.In the time fusion process,the error of time fusion is greatly reduced by the best matching function to ensure the synchronization of the time axis.In the process of spatial fusion,the visual coordinate system is selected as the reference coordinate system,and the haptic coordinate system is positioned according to the imaging of the haptic device to achieve the consistency of the spatial coordinate system.According to the laws of human motion and some assumptions of human anatomy,a method of reconstructing the 3D gait skeleton length model by visual tactile fusion using monocular view in monocular vision is proposed.In the geometric analysis process of the method,the contour features are used to effectively extract the thigh bone of the human body,which solves the problem that the shoulder joint and the hip joint are difficult to measure due to occlusion,and provides the physical length of the double hip and the shoulder for the gait recognition data source and Its self-rotating motion parameters.The personality of the model between different individuals is preserved,which lays the foundation for the final human gait identification.According to the movement law of the human body,the body parts in different motion states are divided,and the human gait bone quality model is established.Taking the supporting foot as the starting point,the acceleration of the centroid of each part in the model is analyzed according to the kinematics transfer process.Finally,the mathematical model of the mass and the supporting force of each part is established by the formation principle of the supporting force.Through the model,the mass distribution and centroid distance of the human can be solved according to the detected kinematic data and mechanical data,and thus applied to personal feature identification and action prediction.