Image database constructing method and device using the same

Patent metadata

Field	Value
Publication number	US-9508159-B2
Application number	US-201414327681-A
Country	US
Kind code	B2
Filing date	Jul 10, 2014
Priority date	Feb 14, 2014
Publication date	Nov 29, 2016
Grant date	Nov 29, 2016

Abstract

Disclosed is a device for constructing image database, including: an image input unit for receiving an image; and a building identification unit configured to map the image to a three-dimensional map based on photographing information of the image, to project direction vectors toward respective locations of the image that are mapped to the three-dimensional map from a photographing position of the image, and to distinguish buildings from the image based on whether the respective direction vectors collide or not.

First claim

What is claimed is:

1. An image database constructing device for constructing image database, comprising:
an image input unit for receiving a photographed image with photographing information of the image; and
a building identification unit configured to:
construct a three-dimensional sphere from at least one photographed image,
identify a spot corresponding to a photographing position of the photographed image in a three-dimensional map, and set the three-dimensional sphere in the spot of the three-dimensional map based on the photographing information of the photographed image,
map the photographed image to the three-dimensional map based on the three-dimensional sphere and the photographing information of the photographed image to create a three-dimensionally mapped image,
project direction vectors toward respective locations of the three-dimensionally mapped image from the photographing position of the three-dimensionally mapped image,
determine whether the direction vectors collide with a location in the three-dimensionally mapped image,
assign an identifier to a location that is collided with the direction vectors and distinguish a building from the location in the three-dimensionally mapped image based on the identifier, and
determine, if a first direction vector projected to a first location of the photographed image collides with a three-dimensional object of the three-dimensional map, the first location as a building area.

2. The device of claim 1 , wherein the building identification unit is further configured to assign, if the first direction vector collides with the three-dimensional map including building information, the building information corresponding to the collided location to the first location.

3. The device of claim 2 , wherein the building identification unit is further configured to:
extract the building area from the mapped image based on the determination, and
partition partitions the building area into several buildings based on the building information of the respective locations included in the building area.

4. The device of claim 3 , wherein the building identification unit is further configured to:
group locations assigned with a first building identifier in the building area, and
distinguish the locations as a first building.

5. The device of claim 3 , further comprising
a building information mapper configured to map building-specific information of the buildings to the building area that is partitioned by the building identification unit.

6. The device of claim 2 , wherein the building information is a building identifier of the three-dimensional building with which the first direction vector collides.

7. The device of claim 1 , wherein the building identification unit is further configured to project the direction vectors into respective pixels of the photographed image.

8. The device of claim 1 , wherein the photographing information of the photographed image includes photographing position information and azimuth information.

9. An image database constructing method using an image database constructing device, comprising:
receiving a photographed image with photographing information of the image;
construct a three-dimensional sphere from at least one photographed image,
identifying a spot corresponding to a photographing position of the photographed image in a three-dimensional map, and setting the three-dimensional sphere in the spot of the three-dimensional map based on the photographing information of the photographed image;
mapping the photographed image into the three-dimensional map based on the photographing information of the photographed image and the three-dimensional sphere to create a three-dimensionally mapped image;
projecting direction vectors from the photographing position of the three-dimensionally mapped image to respective locations of the three-dimensionally mapped image;
determining whether the direction vectors collide with a location in the three-dimensionally mapped image;
assigning an identifier to a location that is collided with the direction vectors; and
distinguishing a building from the location in the three-dimensionally mapped image based on the identifier,
determining, if a first direction vector projected to a first location of the photographed image collides with a three-dimensional object of the three-dimensional map, the first location as a building area.

10. The method of claim 9 , wherein distinguishing the building from the location in the three-dimensionally mapped image comprises determining, if a first direction vector projected to a first location of the photographed image collides with a three-dimensional object, the first location as a building area and determining if a second direction vector projected to a second location of the image does not collide with the three-dimensional object, the second location as a non-building area.

11. The method of claim 10 , wherein the distinguishing the building from the location in the three-dimensionally mapped image assigns, if the first direction vector collides with a first building of the three-dimensional map, a building identifier of the first building to the first location and distinguishes locations assigned with the building identifier of the first building as the first building, and the three-dimensional map is a map in which the building identifiers of the corresponding buildings are mapped to every building.

12. The method of claim 11 , further comprising
mapping building-specific information of the first building to an area that is distinguished as the first building in the image.

13. The method of claim 9 , wherein the projecting the direction vectors projects the direction vectors to respective pixels of the photographed image.

Assignees

• Kakao Corp

Inventors

• Kim Keetae
• Jung Won Jo
• Park Younguk

Classifications

• G06T7/60 Primary

  Analysis of geometric attributes · CPC title

• G06K9/00671

  Physics · mapped topic

• G06T2207/30184

  Infrastructure · CPC title

• G06V20/20 Primary

  in augmented reality scenes · CPC title