Point cloud data Processing learning notes

Three-dimensional computational vision research includes:

　　Three-dimensional matching

Multi-View three-dimensional reconstruction

SLAM

Target recognition

Shape detection and classification

Semantic classification

Stereo vision and stereo matching ZNCC

Sfm

1, point cloud Filter method ( data preprocessing) :

Bilateral filtering, Gaussian filtering, conditional filtering, pass-through filtering, and random sampling consistent filtering.

Voxelgrid

2, Key points

Iss3d, Harris3d, Narf

Sift3d,

3, characteristics and characteristics of the description

Normal and curvature calculation normalestimation, eigenvalue analysis Eigen-analysis, EGI

PFH, FPFH, 3D Shape Context, Spin Image

4, point cloud matching

ICP, robust ICP, PLICP, MBICP, GICP

Ndt3d, Multil-layer NDT

FPCs, Kfpcs, Sac-ia

Line Segment Matching, ICL

5, point Cloud segmentation

Area growth, Ransac line extraction, K-means, Normalize Cut

6, Slam diagram optimization

G2O, LUM, Elch

Slam methods: ICP, MBICP, IDC, Likehood Field, cross Correlation, NDT

7. Target identification and retrieval

Hausdorff distance calculation (face recognition)

8. Change detection

Eight-fork Tree

Point cloud data Processing learning notes