3d representation reconstruction from images using volumic probability data
A method for generating a three-dimensional representation of a 3D object located in a scene volume. The method comprises obtaining a plurality of source images 1100 recorded by a source camera (Fig. 1, 12a & 12b); executing a skeleton based method to determine a probability 1102 of a voxel belo...
Saved in:
Main Author | |
---|---|
Format | Patent |
Language | English |
Published |
03.03.2021
|
Subjects | |
Online Access | Get full text |
Cover
Loading…
Summary: | A method for generating a three-dimensional representation of a 3D object located in a scene volume. The method comprises obtaining a plurality of source images 1100 recorded by a source camera (Fig. 1, 12a & 12b); executing a skeleton based method to determine a probability 1102 of a voxel belonging to a region of the scene; use a shape-from-silhouette method to generate silhouette images 1108 of the 3D object, segmenting the object from the background image 1107; generating a 3D representation 1111 of the object using the probabilities and silhouettes. The method may further generate a 3D shape volume 1109 including determining a number of background votes for a voxel by counting silhouette images in which said voxel is associated with the image background. The background may be associated with a threshold value to aid determination of background voxels. Further disclosed is a system to perform the method. |
---|---|
Bibliography: | Application Number: GB20190012310 |