A point cloud-based visual segmentation method for interface pore structure of 3D printed concrete

Abstract

The application discloses a kind of 3D printing concrete interface pore structure visual segmentation methods based on point cloud, comprising the following steps;Step 1: point cloud data acquisition is carried out;Step 2: the point cloud data collected is preprocessed;Step 3: point cloud data set is divided into training set, verification set and test set;Step 4: point cloud data sampling is carried out for the point cloud data set training set of step 3 to obtain local relationship;Step 5: using multi-layer perception mechanism constructs the feature detail extraction network in 3D printing concrete interface pore structure visual segmentation network model PointESamba, carries out detail extraction to the feature of the sampled point cloud data of pore structure;Step 6: by adaptive sorting algorithm, the point cloud disorder is optimized;Step 7: the most adaptive sorting is obtained using the sorting algorithm of step 6, the global feature is obtained using Mamba module, and the features extracted in step 5 are further segmented and identified to complete the construction of PointESamba.The application enhances the segmentation accuracy and ability of point cloud to pore structure.

Images