SYSTEM AND METHOD FOR FILTERING A POINT CLOUD

• US 20090048782A1
• Filed: 01/15/2008
• Published: 02/19/2009
• Est. Priority Date: 08/17/2007
• Status: Active Grant

• Current Assignee: Hong Fu Jin Precision Industry (Shenzhen) Co., Ltd. (Hon Hai Precision Industry Co., Ltd.), Hon Hai Precision Industry Co., Ltd.
• Original Assignee: Hong Fu Jin Precision Industry (Shenzhen) Co., Ltd. (Hon Hai Precision Industry Co., Ltd.), Hon Hai Precision Industry Co., Ltd.
• Inventors: Chang, Chih-Kuang, Fu, Shan-Yang
• Granted Patent: US 7,925,110 B2
• Time in Patent Office: Days
• US Class Current: 702/5
• CPC Class Codes: G06T 17/00 Three dimensional [3D] modelling, e.g. data description of 3D objects

Abstract

A method for filtering a point cloud is provided. The method includes: (a) acquiring a point cloud of an object from a point cloud obtaining device; (b) establishing a topological structure for the point cloud; (c) selecting a maiden point from the point cloud as a selected point; (d) searching a plurality of points which are near to the selected point from the point cloud according to the topological structure as near points of the selected point; (e) determining whether the selected point is a noise point by comparing coordinate values of the selected point and coordinate values of the near points; (f) deleting or smoothing the noise point from the point cloud; and repeating steps from (c) to (f), until all points in the point cloud have been selected. A related system is also provided.

Claims

1. A system for filtering a point cloud, the system comprising an application server connected to a point cloud obtaining device, the application server comprising:

a point cloud acquiring module configured for acquiring a point cloud of an object from the point cloud obtaining device;

a topological structure establishing module configured for establishing a topological structure for the point cloud to make points of the point cloud confined in a plurality of related cubical grids;

a point selecting module configured for selecting a maiden point from the point cloud as a selected point;

a near points searching module configured for searching a plurality of points which are near to the selected point from the point cloud according to the topological structure as near points of the selected point;

a noise point determining module configured for determining whether the selected point is a noise point by comparing coordinate values of the selected point with coordinate values of the near points; and

a noise point filtering module configured for deleting or smoothing the noise point from the point cloud.

2. The system as described in claim 1, wherein the near points of the selected point are in the cubical grid which confines the selected point and/or in the cubical grids adjacent to the cubical grid which confines the selected point.

3. The system as described in claim 1, wherein the application server further comprises a judging module configured for judging whether all points in the point cloud have been selected.

4. The system as described in claim 1, wherein the topological structure establishing module establishes the topological structure for the point cloud by way of:

identifying an outermost point on each surface edge of the point cloud by analyzing coordinates values of the points in the point cloud;

determining a cubical figure that confine the point cloud, and retrieving a length, a width and a height of the cubical figure; and

mapping the cubical figure into a plurality of cubical grids according to the length, the width, the height of the cubical figure and an axes interval of the grid that is preconfigured by a user.

5. The system as described in claim 2, wherein the noise point determining module determines whether the selected point is a noise point by way of:

arranging x-coordinate values, y-coordinate values and z-coordinate values of the selected point and the near points into three arrays in an ascending sequence or in a descending sequence respectively;

determining whether at least one of the x-coordinate value, the y-coordinate value and the z-coordinate value of the selected point is at the ends of the corresponding array;

ascertaining that the selected point is a normal point if none of the x-coordinate value, the y-coordinate value and the z-coordinate value of the selected point is at the ends of the corresponding array;

or

ascertaining that the selected point is a noise point if at least one of the x-coordinate value, the y-coordinate value and the z-coordinate value of the selected point is at the ends of the corresponding array.

6. The system as described in claim 5, wherein the noise point filtering module smoothes the noise point if only one value of the x-coordinate value, the y-coordinate value and the z-coordinate value of the selected point is at the ends of the corresponding array, or deletes the noise point if at least two values of the x-coordinate value, the y-coordinate value and the z-coordinate value of the selected point are at the ends of the corresponding array.

7. A computer-enabled method for filtering a point cloud, the method comprising steps of:

(a) acquiring a point cloud of an object from a point cloud obtaining device;

(b) establishing a topological structure for the point cloud;

(c) selecting a maiden point from the point cloud as a selected point;

(d) searching a plurality of points which are near to the selected point from the point cloud according to the topological structure as near points of the selected point;

(e) determining whether the selected point is a noise point by comparing coordinate values of the selected point and coordinate values of the near points;

(f) deleting or smoothing the noise point from the point cloud; and

repeating steps from (c) to (f), until all points in the point cloud have been selected.

8. The method as described in claim 7, wherein the step (b) comprises steps of:

(b1) identifying an outermost point on each surface edge of the point cloud by analyzing coordinates values of the points in the point cloud, and determining a cubical figure that confine the point cloud;

(b2) retrieving a length, a width and a height of the cubical figure;

(b3) computing an axes quantity in surfaces of the cubical figure according to the length, the width, the height of the cubical figure and an axes interval of the grid preconfigured by a user;

(b4) mapping the cubical figure into a plurality of cubical grids according to the axes quantity of each surface of the cubical figure;

(b5) configuring serial numbers for all cubical grids;

(b6) relating one of the cubical grids with twenty-six cubical grids which are adjacent to the cubical grid in three-dimensional space by saving the serial numbers of the twenty-seven cubical grids into a table; and

repeating step (b6) until all cubical grids have been related.

9. The method as described in claim 8, wherein the step (d) comprises steps of:

(d1) obtaining a serial number of the cubical grid which confines the selected point, and searching twenty-six cubical grids which are adjacent to the cubical grid according to the serial number;

(d2) judging whether quantity of the points confined in the searched cubical grids is more than a predetermined value;

(d3) searching other cubical grids which are adjacent to one of the searched cubical grids until the quantity of the points is more than the predetermined value; and

(d4) computing a distance between the selected point and each of the points that is confined in the searched cubical grids to obtain the near points of the selected point according to the distances.

10. The method as described in claim 7, wherein the step (e) comprises steps of:

(e1) respectively arranging x-coordinate values, y-coordinate values and z-coordinate values of the selected point and the near points into three arrays in an ascending sequence or in a descending sequence;

(e2) determining whether at least one of the x-coordinate value, y-coordinate value and z-coordinate value of the selected point is at the ends of the corresponding array; and

(e3) ascertaining the selected point is a noise point if at least one is at the end of the corresponding array;

or

(e4) ascertaining the selected point is a normal point if none of the x-coordinate value, y-coordinate value and z-coordinate value of the selected point is at the ends of the corresponding array.

11. The method as described in claim 10, wherein step (f) comprises steps of:

smoothing the noise point if only one of the x-coordinate value, y-coordinate value and z-coordinate value of the selected point is at the ends of the corresponding array;

or

deleting the noise point if at least two values of the x-coordinate value, y-coordinate value and z-coordinate value of the selected point are at the ends of the corresponding array.

12. The method as described in claim 11, wherein the means of smoothing the selected point can be using a mean value method or an intermediate value method.