###2020 Phoenix project #Dependencies
- Numpy
- Numpy-stl
- py-goicp
- (https://github.com/aalavandhaann/go-icp_cython). Download from pip or you can manually compile.
- Issue on mac pip install requires manual compiling.
Included are scripts pertaining to realignment.
An example run looks like 'python protectRunner.py meshes/AnkleMesh_1_L_Input.stl AnkleAtlasL.stl'
- meshes/AnkleMesh_1_L_Input.stl is the input file
- AnkleAtlasL.stl is the current atlas.
- The output is saved to meshes/AnkleMesh_1_L_Input_reoriented.stl
#GO Icp Params
- mseUnscaled: default = .04
- This is the threshold for defining correct alignment. A lower value means longer runtime but more precise output
- pruneAmount: default = 200
- Number of data points to use in atlas. Lower value decreases runtime but could affect accuracy.
- trimFraction: default = .15
- Percentage of atlas data points to be considered outliers during go-icp. Useful if region of atlas does not exist in input (input scan has shorter shin length etc.). Setting to 0 can help for many input scans. Small increase in runtime for some examples, for others it is faster.
- dtSize: default = 150
- Used to specify size of closest point framework. Higher value means more precise error measurement but much longer runtime. I would recommend changing after other values and not going beyond 300.