Design And Implementation Of Hand Tracking System Based On Depth Sensor

The question of how machine vision and human interaction can better serve humanity is a major development problem in today's world.Along with the depth sensor technology mature,machine vision is used in the human learning life more widely,for example the sweeping robot such as artificial intelligence into the embodiment of human life.The gesture is a very convenient and efficient means of interaction,the use of auxiliary sensing device like data glove,motion-sensing technology research and development constantly widening the channels of interpersonal interaction,including the research for the identification and tracking of a man's hand.The study and implementation of manual tracking are mainly intended to apply this feature to the home environment,then combine the devices such as smart homes provide the people who are disabled in family life convenience.This paper mainly uses the particle filter algorithm to realize the identification and tracking of naked hands of human body,it is an improvement on the Markov-monte carlo algorithm.The difference from the normal track implementation,the human hand has a relatively uniform color,and the shape changes at different times,so this article will make the color features and depth information relate each other.Based on the shape of the naked hand and the color luminance,a depth sensor is used to track human hands.This system uses the information such as gesture movement speed,position,shape,and the size of the target area of the previous time to predicted the gesture position of present moment.Using the integral diagram in the calculation of the prediction to reduce the difference in moments before and after,and tracking is more efficient.In this paper,the manual tracking of the system is carried out in different light,different gestures,and occlusion,etc.The result of the experiment is that the light condition has a certain effect on the tracking of gesture recognition.This paper attempts to overcome the effects of light by filling holes in the target area,and by integrating graphs when they change at the next moment,when they are occlusion finally the efficiency has been improved,the differences can be tracking.The target feature information that was tracked for the previous time was applied to the current time tracking.The target location of the current time is predicted based on the distance of the target,the difference in the number of pixels of the area and the change of velocity.The results are more reliable.Therefore,the system can be tracked in different state,and the efficiency is higher,combined with the use of gestures in our daily life,it can be divided into different meanings according to the speed of the hand movement speed and the moving trajectory,make the human-computer interaction more convenientby applying this system to the home service robot.