Motion Estimation And Analysis Of Deployable Mechanism Based On Image

The object of the thesis is deployable mechanism built in ADAMS, and according to its structural features and motion characteristics set mark blocks on intermediate state models by VRML, and set left and right viewpoints to build parallel binocular stereo vision model. Then collect image sequences of unfolding process of deployable mechanism by some sampling rate and make motion estimation and analysis on it through the image sequences.In traditional stereo vision field, apply the measurement method based on parallel binocular stereo vision theory to obtaining 3D position information of feature points of mark blocks on the deployable mechanism, and measure and calculate motion track and velocity parameter combining with sample rate.Based on motion model of deployable mechanism and related to Bayesian theory, the paper makes MRF-MAP algorithm Bayesian theory-based used in 3D motion estimation, applies Markov random field to image model and set marking field X motion parameters to describe motion states of feature points. On the basis of consistency of Markov random field and Gibbs distribution, it is changed to problem of minimizing a posteriori energy from maximizing a posteriori probability. Motion track and velocity parameter of feature points set on deployable mechanism are achieved by objective function of MRF-MAP algorithm solved of global-optimization by improving simulated annealing algorithm. It is pilot study and analysis that the thesis does on application of MRF algorithm in the field of motion estimation.For problems in motion estimation simulation of deployable mechanism, the paper has in-depth discussion and analysis in aspects of deployable mechanism motion model, sampling rate of image sequences, camera resolution, feature detection, and presents corresponding improving methods of certain significance and instruction on the study and analysis of practical motion estimation of deployable mechanism.