论文部分内容阅读
已知含有多个三维刚体的场景，在运动前后的二维点对应数据集合，其中可以包含高斯噪声和出格点数据，发展了初始部分匹配的生成－生长技术并运用刚性约束．将上述二维点对应数据集合，分割成多个分别对应于不同刚体运动的二维点对应数据子集，并能分离出所有出格点数据．再利用单刚体运动估计算法就可估计出各个刚体运动参数．实验结果表明了算法的有效性．
It is known that a scene containing a plurality of three-dimensional rigid bodies corresponds to a data set of two-dimensional points before and after the movement, which can include Gaussian noise and outlier data. A generation-growth technique of initial partial matching is developed and a rigid constraint is applied. The above-mentioned two-dimensional point correspondence data set is divided into a plurality of two-dimensional point correspondence data subsets respectively corresponding to different rigid body motions, and all out-point data can be separated. Re-use of a single rigid body motion estimation algorithm can estimate the various rigid body motion parameters. Experimental results show the effectiveness of the algorithm.