Research On Anthropmetrics In Orbital Flight Measured By Markless Videometrics

The development of China manned space engineering requires the analysis of human habits and physical conditions in space environment.Anthropometic parameters in the space environment is the basic data of analysis.In this paper,key techniques of unmarked visual method for anthropometic was researched.Space mission has special requirements for the volume,power consumption and efficient of the device.The technique of this paper is reconstruct the body by binocular camera.Then get the anthropometic parameters like height,shape,occupying space and pose.Experiment shows that this system has advantages of unmarkers,high precision,easy to operate and low power consumption,satisfied the demand of space mission.Aiming at 3-D reconstruction,this paper studied methods go for moving object and static objects.Method goes for moving object aims at moving body,which use binocular camera with a IR projection.Infrared projector random spot increases the texture of target.Binocular camera captures single frame image and make dense stereo matching to obtain depth information.Methods go for static object also use binocular camera with a projection.Unlike the method go for moving object,the projector is encoded structured light.Binocular camera captures multiple-frame images,decode and stereo matching.Precision of this method is significantly higher than single-frame capture,and can be used for standard human model.Full body model needs muti-view collection.Point clouds registration involves camera pose estimation.This paper discussed two different kinds of point registration methods,including registration based on features of point clouds,and coding markerassisted method.The precision of the two methods were discussed through simulation experiment.And their characteristics were analysised.Experiment shows that both methods can achieve a higher precision ideally.Method based on features of point clouds depends on the public of different clouds and initial value.Coding marker-assisted method needs to set up special markers,but does not require initial values.And marker position algorithm shows higher robustness.For the reconstruction of human body,because of the small public of different point-clouds,Coding marker-assisted method were needed.This paper also studied how to get the anthropometric parameters through the human body model.The precision and effectiveness of method were evaluated by simulation experiment and contrast experiment.Heigh and shape of body were directly related to the precision of body point clouds,which were evaluated by reconstruct a standard component.Human occupied space were calculated by minimum enclosing cube build by principal component analysis,compared with the global search within a 1 % error.Human pose estimation was achieved by model matching method.And this paper showed how to build skeleton-driven mathematical human models.Experiments was designed in contrast to moven.This paper researched measuring anthropmetrics in orbital flight by markless videometrics.With the development of algorithms and hardware system,the accuracy and efficiency would be further improved.The application is not only for human factors analysis but also for human-computer interaction and virtual reality.