You're reading the documentation for a development version. For the latest released version, please have a look at master.

Overview

Input

  • Cloud Ptr: a point cloud to be segmented

  • Camera Intrinsic: camera intrinsic parameter used for RGBD segmentation

  • RGB Pointer: RGB image to be segmented

Output

  • preSegmRGB: rgb image before segmentation

  • postSegmDepth: depth image of the scene after segmentation

  • postSegmPCA: image showing PCA of the scene cluster

  • postSegmRGB: rgb image after segmentation

  • results: a vector of segments, each containing a mask image of the size of the scene image, showing which points belong to this segment

Parameters

Pre-process

  • Basic Downsample: Level of downsampling when doing segmentation

Segmentation Parameters

  • Basic Weight Threshold: how much different between the adjacent points to be considered as different segment. Default Value is 50, range is [0,100].

  • Basic Min Size: minimum size of each segment. Default value is 50, range is [0,100].

Post-processing

  • Max Segment Size: Maximum size of the segment, segment greater than this will be removed. Default value is 100, range is [0,100].

  • Min Segment Size: Minimum size of the segment, segment greater than this will be removed. Default value is 0, range is [0,100].

  • Max Distance to Camera: Maximum distance of points in segment to camera, segment with points further than this distance will be removed. Off by default. Distance unit is millimeter

  • Use Nan Removal: remove segment with more than 75% of the points invalid, if turned on.