Personalized Hand Modeling Based On Computer Vision

With the development of virtual reality technology,the natural human-computer interaction has attracted much attention.As an important organ for human-computer interaction,hand has gradually become an important research object,and has been widely used in virtual assembly,virtual training,somatosensory games and other fields.The key technology to achieve immersive interaction is to create a kinematic model which consistent with the real hand.This paper studies the method of initializing the human hand kinematics model using computer vision technology.In this paper,the physiological structure of the hand and the motion characteristics of the joints are analyzed in detail,and the hand mechanism model is established.The kinematics analysis of the mechanism model is carried out by using D-H parameter method,which establishes the theoretical basis for parametric control of position and attitude.The object-oriented programing is used to build a virtual hand model,which provides a program interface for the kinematics modeling of hands.This paper studies the method of gesture segmentation based on complexion model.The ellipse model is used to detect complexion initially in YCbCr color space.In order to eliminate the holes and edge defects in complexion detection,the circular gradient is used to detect the edge of complexion,so as to fill the missing and get more accurate gesture segmentation result.Taking the frontal gesture image with five fingers spread as the research object,and an algorithm of extracting gesture gravity center based on adaptive times of corruption is proposed.This paper improves the detection algorithm of fingertips and finger valleys based on contour distance distribution.The ROIs of five fingers are determined by fingertips and finger valleys.In these regions,and the principal lines of palmprint and the knuckleprint are positioned according to the jump characteristics of the gray gradient value,which reflects the joints position and realizes the personalized human hand kinematics modeling.After several groups of experiments,the method of instantiating human hand kinematics model based on computer vision has good accuracy and good stability for image scaling and rotation.