Tchebichef moment shape descriptor for partial point cloud characterization

Patent metadata

Field	Value
Publication number	US-9934590-B1
Application number	US-201615283853-A
Country	US
Kind code	B1
Filing date	Oct 3, 2016
Priority date	Jun 25, 2015
Publication date	Apr 3, 2018
Grant date	Apr 3, 2018

Abstract

A process and apparatus are provided to characterize low-resolution partial point clouds for object recognition or query. A partial point cloud representation of an object is received. Zero and first order geometric moments of the partial point cloud are computed. A location of a center of a point cloud mass is computed using the geometric moments. A cubic bounding box is generated centered at the location of the mass center of the point cloud, with one side of the box bounding the point cloud at its longest semi-axis. The bounding box is divided into a three dimensional grid. A normalized voxel mass distribution is generated over the three dimensional grid. Tchebichef moments of different orders are calculated with respect to the voxel mass distribution in the grid. Low-order moments are collected to form TMSDs. Similarity is compared between the TMSD of the point cloud with TMSDs of other point clouds.

First claim

What is claimed is:

1. A process for characterizing global shape pattern of low-resolution, partial point clouds, the process comprising:
receiving a partial point cloud representation of an object from a sensor;
computing zero and first order geometric moments of the partial point cloud;
computing a location of a center of a point cloud mass using the zero and first order geometric moments;
generating a bounding box;
dividing the bounding box into a three dimensional grid;
generating a normalized voxel mass distribution over the three dimensional grid;
calculating Tchebichef moments of different orders with respect to the voxel mass distribution in the grid; and
collecting low-order moments to form a one-dimensional numerical vector containing 3D Tchebichef Moment Shape Descriptors (TMSD).

2. The method of claim 1 , further comprising:
comparing the similarity between the TMSD of the point cloud with TMSDs of other point clouds of known classes of shapes for partial point cloud based object recognition or query.

3. The method of claim 2 , wherein comparing the similarity comprises:
a multi-scale nearest neighbor (NN) query.

4. The method of claim 1 , wherein generating a bounding box comprises:
generating a bounding box centered at the location of the center of the point cloud mass.

5. The method of claim 4 , wherein one side of the box bounding the point cloud is at its longest semi-axis.

6. The method of claim 1 , wherein the bounding box is a cubic bounding box.

7. The method of claim 1 , wherein the three dimensional grid is a three equal-dimensional grid.

8. The method of claim 7 , wherein the bounding box is divided into an N×N×N grid, where N is selected from a group consisting of: 16, 32, 64.

9. An apparatus, comprising:
a sensor configured to generate a partial point cloud representation of an object;
a memory in electrical communication with the sensor and configured to store the partial point cloud generated by the sensor;
a processor in electrical communication with the memory; and
program code resident in the memory and configured to be executed by the processor to characterize partial point clouds, the program code further configured to retrieve the partial point cloud representation of an object stored in the memory, compute zero and first order geometric moments of the partial point cloud, compute a location of a center of a point cloud mass using the zero and first order geometric moments, generate a bounding box, divide the bounding box into a three dimensional grid, generate a normalized voxel mass distribution over the three dimensional grid, calculate Tchebichef moments of different orders with respect to the voxel mass distribution in the grid, and collect low-order moments to form a one-dimensional numerical vector containing 3D Tchebichef Moment Shape Descriptors (TMSD).

10. The apparatus of claim 9 , wherein the program code is further configured to:
compare the similarity between the TMSD of the point cloud with TMSDs of other point clouds of known classes of shapes for partial point cloud based object recognition or query.

11. The apparatus of claim 10 , wherein comparing the similarity comprises:
a multi-scale nearest neighbor (NN) query.

12. The apparatus of claim 9 , wherein generating a bounding box comprises:
generating a bounding box centered at the location of the center of the point cloud mass.

13. The apparatus of claim 12 , wherein one side of the box bounding the point cloud is at its longest semi-axis.

14. The apparatus of claim 9 , wherein the bounding box is a cubic bounding box.

15. The apparatus of claim 9 , wherein the three dimensional grid is a three equal-dimensional grid.

16. The apparatus of claim 15 , wherein the bounding box is divided into an N×N×N grid, where N is selected from a group consisting of: 16, 32, 64.

17. A program product, comprising:
a non-transitory computer recordable type medium; and
a program code configured to be executed by a hardware based processor to characterize partial point clouds, the program code further configured to retrieve the partial point cloud representation of an object from a sensor, compute zero and first order geometric moments of the partial point cloud, compute a location of a center of a point cloud mass using the zero and first order geometric moments, generate a bounding box, divide the bounding box into a three dimensional grid, generate a normalized voxel mass distribution over the three dimensional grid, calculate Tchebichef moments of different orders with respect to the voxel mass distribution in the grid, and collect low-order moments to form a one-dimensional numerical vector containing 3D Tchebichef Moment Shape Descriptors (TMSD).

18. The program product of claim 17 , wherein the program code is further configured to:
compare the similarity between the TMSD of the point cloud with TMSDs of other point clouds of known classes of shapes for partial point cloud based object recognition or query.

19. The program product of claim 18 , wherein comparing the similarity comprises:
a multi-scale nearest neighbor (NN) query.

20. The program product of claim 17 , wherein generating a bounding box comprises:
generating a cubic bounding box centered at the location of the center of the point cloud mass,
wherein one side of the box bounding the point cloud is at its longest semi-axis.

Assignees

• Us Air Force
• Us Air Force

Inventors

• Cheng Huaining

Classifications

• G06V10/435

  Computation of moments · CPC title

• G06V10/50

  by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis · CPC title

• G06V10/431

  Frequency domain transformation; Autocorrelation · CPC title

• G06T2207/10028

  Range image; Depth image; 3D point clouds · CPC title

• G06T2207/20021

  Dividing image into blocks, subimages or windows · CPC title