Method for identifying semantic information of indoor scene point cloud component

Abstract

The invention relates to an indoor scene point cloud part semantic information recognition method, and the method specifically comprises the following steps: obtaining three-dimensional laser point cloud data of an unknown semantic category, putting the three-dimensional laser point cloud data into point cloud data of a known category, and obtaining a known-unknown confused point cloud data set; extracting global features of the point cloud data set in the high-dimensional space; dimension reduction processing is carried out on the global features in the high-dimensional space; clustering theclusters obtained after dimension reduction to obtain a cluster set of samples in a low-dimensional space; on the basis of low-dimensional spatial clustering, for sample points in each cluster, a known semantic sample and an unknown semantic sample are found respectively, and semantic assignment is performed on the unknown semantic sample by using category information of the known semantic sample.Compared with the prior art, the method has the advantages that semantic information of a large batch of object point clouds can be obtained only by using a small number of known samples; the methodis low in memory overhead, obviously improves the calculation efficiency, and has low requirements on a computer.

Description

Technical field [0001] The invention relates to the field of surveying and mapping and intelligent robots, and in particular to a method for identifying semantic information of point cloud components in indoor scenes. Background technique [0002] Recognizing the semantic information of the indoor scene point cloud components is to analyze the structure of the point cloud data of each object in the indoor scene collected by the sensor, and assign the category semantic information to each object, that is, the classification of the indoor scene object. This is to understand the three-dimensional The basic task of the scene. It is widely used in technologies such as intelligent robots, unmanned driving, real-time positioning and mapping (SLAM), surveying and mapping, and indoor high-precision mapping. [0003] There are mainly two existing identification methods: [0004] 1. Object recognition based on traditional methods mainly uses a combination of manual shape descriptors and machi...

AI Technical Summary

Problems solved by technology

However, the recognition method of deep learning requires a large amount of sample data for training, which requires a large amount of calculation, and has high requirements for computers and data sets. Moreover, for the phenomenon of point cloud occlusion, dynamic target interference, and object overlapping in real scene measurements, the recognition accuracy is very low. Greatly reduced

Images