Collaborative disparity decomposition
A novel disparity computation technique is presented which comprises multiple orthogonal disparity maps, generated from approximately orthogonal decomposition feature spaces, collaboratively generating a composite disparity map. Using an approximately orthogonal feature set extracted from such feature spaces produces an approximately orthogonal set of disparity maps that can be composited together to produce a final disparity map. Various methods for dimensioning scenes and are presented. One approach extracts the top and bottom vertices of a cuboid, along with the set of lines, whose intersections define such points. Tt then defines a unique box from these two intersections as well as the associated lines. Orthographic projection is then attempted, to recenter the box perspective. This is followed by the extraction of the three-dimensional information that is associated with the box, and finally, the dimensions of the box are computed. The same concepts can apply to hallways, rooms, and any other object.
1. A method comprising:
receiving a plurality of images associated with an object;
defining a plurality of vertices of the object;
matching the plurality of vertices of the object across the plurality of images to extract a three-dimensional vector for each of one or more vertex combinations;
extracting a line segment connecting a first vertex of the plurality of vertices of the object and a second vertex of the plurality of vertices of the object;
projecting the first vertex of the object and its associated edges to a canonical position;
projecting the three-dimensional vector associated with the second vertex to the canonical position;
extracting a z-component of one or more sides defining the first vertex of the object;
projecting the line segment to the canonical position;
extracting one or more angular relationships between the one or more sides and the line segment of the object; and
dimensioning the object in accordance with at least the one or more angular relationships and the line segment.
2. The method of claim 1 , further comprising receiving an input identifying the plurality of vertices of the object.
3. The method of claim 1 , further comprising matching vanishing lines from one or more images of the plurality of images to identify the first vertex.
4. The method of claim 3 , further comprising matching vanishing lines from one or more images of the plurality of images to define the second vertex.
5. The method of claim 1 , wherein the object is an irregular shaped box, an irregular shaped room, or an irregular shaped object.
6. The method of claim 1 , further comprising defining a composite depth map to identify the first vertex.
7. The method of claim 6 , further comprising the steps of:
defining a first complementary depth map having a first dimension;
defining a second complementary depth map having a second dimension, the second dimension being substantially orthogonal to the first dimension; and
combining the first complementary depth map and second complementary depth map to provide the composite depth map.
8. The method of claim 1 , wherein the plurality of images includes a right image of a stereo pair and a left image of the stereo pair.
9. The method of claim 1 , wherein the plurality of images includes three or more images.
10. The method of claim 1 , further comprising the step of modifying the dimensioned object in accordance with a global model.
11. The method of claim 1 , further comprising the steps of:
projecting a structured light pattern on the object;
determining one or more discontinuities in the structured light pattern; and
determining one or more edges of the object in accordance with the one or more discontinuities.
12. A system comprising:
an imaging apparatus configured to image an object; and
a processor configured to:
receive a plurality of images associated with the object;
define a plurality of vertices of the object;
match the plurality of vertices of the object across the plurality of images to extract a three-dimensional vector for each of one or more vertex combinations;
extract a line segment connecting a first vertex of the plurality of vertices of the object and a second vertex of the plurality of vertices of the object;
project the first vertex of the object and its associated edges to a canonical position;
project the three-dimensional vector associated with the second vertex to the canonical position;
extract a z-component of one or more sides defining the first vertex of the object;
project the line segment to the canonical position;
extract one or more angular relationships between the one or more sides and the line segment of the object; and
dimension the object in accordance with at least the one or more angular relationships and the line segment.
13. The system of claim 12 , wherein the imaging apparatus is a stereo imaging apparatus.
14. The system of claim 13 , wherein the imaging apparatus is configured to obtain a left image and a right image imaged by the stereo imaging apparatus.
15. The system of claim 12 , wherein the object is an irregular shaped box, an irregular shaped room, or an irregular shaped object.
16. The system of claim 12 , further comprising an input device configured for receiving an input indicating a location of the plurality of vertices of the object.
17. The system of claim 12 , wherein the processor is further configured to match vanishing lines to identify the plurality of vertices of the object.
18. The system of claim 12 , further comprising a structured light projector configured to project a structured light pattern on the object.
19. The system of claim 18 , wherein the processor is further configured to:
determine one or more discontinuities in the structured light pattern; and
determine one or more edges of the object in accordance with the one or more discontinuities.
20. A non-transitory computer readable storage medium having a computer program stored thereon, the computer program causing a computer system to perform:
receiving a plurality of images associated with an object;
defining a plurality of vertices of the object;
matching the plurality of vertices of the object across the plurality of images to extract a three-dimensional vector for each of one or more vertex combinations;
extracting a line segment connecting a first vertex of the plurality of vertices of the object and a second vertex of the plurality of vertices of the object;
projecting the first vertex of the object and its associated edges to a canonical position;
projecting the three-dimensional vector associated with the second vertex to the canonical position;
extracting a z-component of one or more sides defining the first vertex of the object;
projecting the line segment to the canonical position;
extracting one or more angular relationships between the one or more sides and the line segment of the object; and
dimensioning the object in accordance with at least the one or more angular relationships and the line segment.