Functional modeling method for indoor scene

Abstract

The invention provides a functional modeling method for an indoor scene. The functional modeling method includes the first step of extracting the pose of a depth camera, integrating the pose of the depth camera into an ICP algorithm and carrying out real-time registration reestablishment on three-dimensional surfaces according to the ICP algorithm, the second step of carrying out segmenting and clustering to realize semantic segmentation of the scene, the third step of extracting point-cloud sets with the area larger than a preset threshold value, the fourth step of judging whether point-cloud sets perpendicular to the ground exist, the fifth step of extracting a handle, the sixth step of pulling out the handle, recording the opening process, the closing process and the internal structure and carrying out functional labeling, the seventh step of judging whether point-cloud sets parallel to the ground exist, the eighth step of carrying out Euclid clustering and segmenting to obtain a plurality of individual objects, the ninth step of moving away the individual objects, recording blocked parts and carrying out functional labeling, and the tenth step of carrying out summarizing to obtain a functional model of the indoor scene. According to the method, functional operation can be conducted on the parts with labels, so interactive experience of users is greatly promoted.

Description

technical field [0001] The invention relates to three-dimensional modeling, in particular to a functional modeling method for indoor scenes. Background technique [0002] The digital information of the objective environment can be truly reproduced in the computer, and the environmental information can be processed and analyzed conveniently. The digital representation of the three-dimensional scene information enables the real scene to fully display the details from different perspectives. In recent years, the rapid development of a variety of 3D measurement equipment has also provided more means for 3D scene reconstruction. [0003] There are two traditional methods for 3D modeling of indoor scenes: one is to fix scene data acquisition devices, such as depth cameras, laser radars, binocular vision cameras, etc., on a certain part of the mobile robot, and then the mobile robot Navigate in the indoor environment to complete the scanning work; the other is to directly hold the...

AI Technical Summary

Problems solved by technology

[0005] (1) The existing 3D modeling methods for indoor scenes are affected by factors such as relatively cluttered indoor environments and serious occlusions, making it difficult to obtain a relatively complete 3D model, and generally only reconstruct the surfaces of objects in indoor scenes. It is not possible to reconstruct the interior of objects (such as the interior of cabinets);

[0006] (2) The existing methods and technologies for 3D modeling of indoor scenes have not realized the functional reconstruction of the scene, and the reconstruction of the existing methods has lost the functional information of meaningful objects in the scene

Images