Deep Learning Point Cloud Classification Algorithm Considering Local Spatial Features

In the process of feature extraction of point cloud data, most of the existing methods ignore the rich spatial geometric information in point cloud data, resulting in poor classification accuracy To solve this problem, this paper proposes a point cloud depth learning classification algorithm considering the geometric characteristics of local space The LANet network model is constructed, and a local spatial feature network is proposed to learn the rich spatial geometric relationships in the point cloud data, and the attention mechanism is added to the network to learn the weight of the feature channel on the classification, so as to further improve the accuracy of point cloud classification The simulation experiment on the ModelNet40 classification dataset shows that the local spatial feature learning network can extract complete spatial geometric information, and the attention mechanism can optimize the classification network. Compared with the current mainstream networks, the proposed method can effectively learn point cloud data and help improve the accuracy of point cloud classification.