LIS Segmentation

LIS Segmentation - prepare point cloud data for object based analysis and classification

The package contains a comprehensive set of point cloud segmentation and region growing tools, each dedicated to specialized tasks. Applications include the segmentation of point clouds by robust plane or line fitting and the segmentation of ground, building and vegetation points.


Tool: Block Segmentation

Features
• point cloud segmentation by robust plane fitting
• data is processed in 2D or 3D tiles
• user specified number of planes to fit within each tile
• output attributes: region id, segment id, normal vector
Application
• point cloud segmentation
• normal vector computation
• object based analysis/classification

Tool: Clump Segmentation

Features
• point cloud segmentation by robust plane fitting
• data is processed in clumps, identified by an attribute
• model points can be specified by a seed attribute
• output attributes: segment id, segment size, normal vector
Application
• point cloud segmentation (esp. building roofs)
• normal vector computation
• object based analysis/classification

Tool: Create Ground Seeds

Features
• detection of ground seed points by iterative TIN densification
• constraining factors: normal tolerance, valid segment
• all valid segments that include a ground seed can be flagged as ground
Application
• region growing
• point cloud classification

Tool: Create Segment Seeds

Features
• creation of seed points on point cloud segments based on attribute values
• seed points are determined by searching the minimum, maximum or median attribute value per segment
Application
• region growing
• object based analysis/classification

Tool: Density-based Spatial Clustering

Features
• data clustering to group points that are closely packed together, marking points in low density regions as outliers (DBSCAN)
• clustering can be constrained by an attribute
• clusters can be filter based on size
• outputs: cluster id, cluster size
Application
• data clustering
• region growing
• object based analysis/classification

Tool: Dijkstra Growing

Features
• point cloud segmentation based on Dijkstra region growing from seed points
• iterative region growing to handle data gaps
• outputs: segment id, parent id, path distance, iteration, distance difference
Application
• point cloud segmentation (esp. trees)
• object based analysis/classification

Tool: Hybrid Segmentation

Features
• point cloud segmentation by distance-weighted cut and value aggregation
• input attributes: normal vector, 3D density, eigenvalues, planarity; optionally RGB
• outputs: plane offset, segment id
Application
• point cloud segmentation
• object based analysis/classification

Tool: Invert Normals

Features
• inversion of point cloud normals
Application
• inverting normal direction

Tool: Orient Normals

Features
• re-orient normals of a point cloud in a consistent way
• normal orientation is propagated from neighbor to neighbor
• constraints: search radius and point density
Application
• correction of normal vectors

Tool: Region Growing

Features
• region growing based on attribute value similarity and nearest neighbor search
• up to three attributes, each with its own value tolerance, and/or normal vector similarity
• random seed or seed attribute
• 2D and 3D nearest neighbor search (knn/radius)
• outputs: segment id, segment size, segment class
Application
• region growing
• object based analysis/classification

Tool: Region Growing [interactive]

Features
• interactive version of the Region Growing tool
• seed point is given by user (mouse-click)
Application
• region growing
• object based analysis/classification

Tool: Segment Centroids

Features
• condense segments to their centroids
• optional: output of mean attribute value per segment
Application
• point cloud segmentation
• object based analysis/classification

Tool: Segment Thinning

Features
• thinning of points on point cloud segments by 3D block filtering
• filter methods: lowest, highest, nearest, mean, median, z-slice mean, min(attribute), max(attribute), centroid
Application
• point cloud thinning

Tool: Segmentation by Lines

Features
• point cloud segmentation by robust line fitting
• threshold attribute constrains which points are processed
• 2D or 3D line fitting
• outputs: segment id, segment size, direction vector, slope, aspect
Application
• point cloud segmentation
• direction vector computation
• object based analysis/classification

Tool: Segmentation by Plane Growing

Features
• point cloud segmentation by robust plane fitting and plane growing
• model points can be specified by a seed attribute, otherwise planarity is used
• radius search or k nearest neighbors
• output attributes: segment id, segment size, normal vector, slope, aspect, quality of fit
Application
• point cloud segmentation
• normal vector computation
• object based analysis/classification

Tool: Segmentation by Planes

Features
• point cloud segmentation by robust plane fitting
• radius search or k nearest neighbors
• output attributes: segment id, segment size, normal vector, slope, aspect, quality of fit
Application
• point cloud segmentation
• normal vector computation
• object based analysis/classification

Tool: Stepwise Segmentation

Features
• point cloud segmentation by robust plane fitting
• optimized for objects with well-defined edges
• model points can be specified by a seed attribute
• radius search
• output attributes: region id, segment id, segment size, normal vector
Application
• point cloud segmentation (eps. buildings and building facades)
• normal vector computation
• object based analysis/classification

Back to LIS