| abstract |
The invention discloses a laser scanning SLAM indoor three-dimensional point cloud quality evaluation method based on deep learning, which includes: S1, obtaining high-quality point clouds through a laser scanning SLAM device; S2, degrading the high-quality point clouds to obtain simulated point clouds ; S3. Perform trajectory measurement analysis on the simulated point cloud; S4. Extract the plane from the high-quality point cloud and the simulated point cloud, perform local consistency noise analysis and geometric rule analysis on the plane, and quantify the quality of the point cloud; S5. Segment the simulated point cloud to obtain point cloud blocks; S6, normalize the point cloud blocks and input them into the PointNet++ neural network for model training to obtain a network model; S7, perform point cloud quality on the point cloud to be evaluated through step S4 Analyze to obtain the point cloud quality level value; S8. Predict the point cloud to be evaluated by the neural network model obtained in step S6, and judge that the point cloud belongs to high-quality point cloud or degraded point cloud. The invention proposes a method for quantifying the point cloud quality, and establishes a classification standard and a framework for evaluating indoor three-dimensional point cloud models under the SLAM system. |