Depth Mapping Using the Hierarchical Reconstruction of Multiple Sequence

摘要

One method was proposed for recovering structure model of the object by using monocular vision equipment. The fundamental idea can be summed up as sequential reconstruction under hierarchical method. First of all, by dividing the single observation path into multiple, more details of structure can be observed. After outliers removing, these sequences can be aligned with the referenced one which can be specified by users. Finally, after surface fitting, depth map of the object can be densely retrieved. The experiment shows that the depth map can be retrieved much densely than traditional method from the same dataset.