Overview

Input

  • Model Mesh: a point cloud or polygon mesh, representing the model

  • Scene Cloud: scene point cloud

  • Hypothesis: a vector of poses for each occurrence of model in scene

Output

  • poses: a vector valid poses after verification, will be sorted if stacking mode is used

Parameters

  • Confidence: the confidence of a pose to be valid. Default is 50, range is [0,100]

  • Depth Mode Parameters

    • Occluded Part Tolerance: threshold such that poses with more than this part occluded should be removed. Default is 50, range is [0,100]

    • Error Tolerance: how much error points a pose can have. Higher the tolerance, more inaccurate result will be kept. Default is 50, range is [0,100]

    • Overlap Ratio: controls the penally low overlap between model and scene. Default is 50, range is [0,100]

  • Edge Mode Parameters

    • Camera Intrinsic Parameter: camera intrinsic parameter, used for edge extract

    • RGB image: RGB image of the scene

    • Smoothness: level of smoothness applied to scene edge. Default is 50, range is [0,100]

    • Canny Low Factor: lower threshold of canny filter

    • Canny High Factor: upper threshold of canny filter

    • Angle Threshold: angle threshold for smoothing the extracted edges

    • Contrast threshold: threshold for filtering out edges with low difference in depth

    • Search Radius: radius used for edge pixel clustering

  • Stacking Mode Parameters

    • FOV: field of view for viewing camera, greater value means more poses will be seen, and then kept.