Construction method of three-dimensional map with character semantic information and processing equipment

Abstract

The invention provides a construction method of a three-dimensional map with character semantic information and processing equipment. The construction method and the processing equipment are used forcreating the three-dimensional map with the character semantic information. The construction method of the three-dimensional map with the character semantic information comprises the steps that a processing device obtains an image sequence obtained by shooting a to-be-reconstructed scene by a camera under different poses of the to-be-reconstructed scene; on the basis of the image sequence, the processing equipment calculates to obtain the three-dimensional structure information of the character and the character pose information of the character of the to-be-reconstructed scene through the character reconstruction thread; the processing equipment calculates three-dimensional environment information of the scene to be reconstructed through a geometric reconstruction thread on the basis of the image sequence; and the processing equipment imports the three-dimensional structure information and the figure pose information into the three-dimensional environment information to obtain a three-dimensional map corresponding to the scene to be reconstructed.

Description

technical field [0001] The present application relates to the field of data processing, in particular to a method for constructing a three-dimensional map with character semantic information and a processing device. Background technique [0002] In the application of 3D maps, the application of semantic 3D maps can provide more functional options than using geometric constraints alone by adding rich semantic information to 3D maps. Semantic 3D maps provide a powerful semantic information database. The database performs information query for the three-dimensional map, so in a specific scene application, the robot can know the attributes of the objects in the scene, so that it can respond, such as avoiding touch or interaction. [0003] In the prior art, the semantic 3D map specifically recognizes the positional objects in the scene and assigns them semantic labels, so that when using the semantic 3D map, a corresponding response can be made according to the attributes of the ...

AI Technical Summary

Problems solved by technology

[0004] However, in further practical applications, it is found that the existing semantic 3D maps are difficult to establish a topologically consistent geometric model for objects with dynamic changes. limit the intelligent application of robots

Images