Method and system to segment depth images and to detect shapes in three-dimensionally acquired data
View Patent ↗A method and system analyzes data acquired by image systems to more rapidly identify objects of interest in the data. In one embodiment, z-depth data are segmented such that neighboring image pixels having similar z-depths are given a common label. Blobs, or groups of pixels with a same label, may be defined to correspond to different objects. Blobs preferably are modeled as primitives to more rapidly identify objects in the acquired image. In some embodiments, a modified connected component analysis is carried out where image pixels are pre-grouped into regions of different depth values preferably using a depth value histogram. The histogram is divided into regions and image cluster centers are determined. A depth group value image containing blobs is obtained, with each pixel being assigned to one of the depth groups.
1. A computing system comprising:
a processor; and
computer storage memory having computer-executable instructions stored thereon which, when executed by the processor, implement a method of three-dimensional (“3D”) image analysis, the method comprising:
receiving 3D image data comprising a plurality of points having 3D coordinate information;
pre-grouping points into sub-regions based upon z-depth values by comparing a first z-depth value for a given pixel with a second z-depth value for a neighboring pixel to determine whether a difference between the first z-depth value and the second z-depth value exceeds a threshold that is used to determine whether the given pixel and the neighboring pixel are assigned the same sub-region;
grouping at least some points within a sub-region to form a segment corresponding to a real-world object of interest;
calculating an image cluster center for the segment; and
modeling the segment as a geometric primitive shape.
2. The system of claim 1 , wherein the real-world object of interest is a human hand.
3. The system of claim 1 , wherein the real-world object of interest is a person's body part.
4. The system of claim 1 , wherein the real-world object of interest is a person.
5. The system of claim 1 , wherein the image cluster center has an assigned depth value that is an average of depth values assigned to points forming the segment.
6. The system of claim 1 , wherein the method further comprises using a time-of-flight 3D camera to capture the 3D image data.
7. One or more computer storage hardware devices having computer-executable instructions embodied thereon that when executed by a computing device perform a method of three-dimensional (“3D”) image analysis, the method comprising:
receiving 3D image data comprising a plurality of points having 3D coordinate information;
pre-grouping points into sub-regions by comparing a first z-depth value for a given pixel with a second z-depth value for a neighboring pixel to determine whether a difference between the first z-depth value and the second z-depth value exceeds a threshold that is used to determine whether the given pixel and the neighboring pixel are assigned the same sub-region;
grouping at least some points within a sub-region to form a segment corresponding to a real-world object of interest;
calculating an image cluster center for the segment; and
modeling the segment as a plurality of geometric primitive shapes.
8. The system of claim 7 , wherein the sub-regions correspond to an object described by the 3D image data.
9. The system of claim 8 , wherein the image cluster center has a depth value equal to an average of depth values of points forming the segment.
10. The system of claim 7 , wherein the real-world object of interest is a person.
11. The system of claim 7 , where in the real-world object of interest corresponds to part of the person.
12. The system of claim 7 , wherein grouping together points into the segment is based upon each point's z-depth value.
13. The system of claim 7 , wherein the sub-regions correspond to a predefined object that is first recognized from the 3D image data.
14. A method of object analysis using three-dimensional image data, the method comprising:
receiving 3D image data comprising a plurality of points having 3D coordinate information;
pre-grouping points into sub-regions based upon a histogram record of z-depth values;
grouping at least some points within a sub-region to form a segment corresponding to a real-world object;
calculating an image cluster center for the segment; and
modeling the segment as a plurality of geometric primitive shapes.
15. The method of claim 14 , wherein the geometric primitive shapes correspond to a 3D object.
16. The method of claim 14 , wherein grouping together at least some points is done based upon their z-depth values and the points are within a number k of neighboring pixels.
17. The method of claim 16 , wherein k is selected from a group consisting of k=4 and k=8.
18. The method of claim 14 , wherein the image center has a depth value equal to an average of depth values of points forming the segment.
19. The method of claim 14 , wherein the 3D image data is received from one or more of an intensity-based system, a time-of-flight system, and a phase-shift z-depth measuring system.
20. The method of claim 14 , wherein the 3D image data is received from one or more of an intensity-based system, a time-of-flight system, and a phase-shift z-depth measuring system.