A Point Cloud Representation Using Plane-Space-Local-Area-Color-Consistency

In order to avoid meshing the dynamic three-dimensional model and mapping texture during the reconstruction of the dynamic object,and then acquire higher reconstruction speed as well as better surface appearance,we present a reconstruction method for dynamic object based on the fact of the plane-space-local-area-color-consistency.First,image segmentation for the every video frame is implemented and then the watershed contours are extracted.Next,the sparse point cloud is computed using the shape-from-silhouettes.Finally,the compact and colored point cloud is attained efficiently based on the plane-space-local-area-color-consistency.Experimental results demonstrate that the new approach can attain the promise.