wbhu
project point cloud to image
hello, I am very interested in your wonderful work! and I am reading your code recently, but some codes I can't understand, I hope you can help.
from ScanNetCross.py line58-59:
p[0] = (p[0] * self.intricsic[0][0]) / p[2] + self.intricsic[0][2]
p[1] = (p[1] * self.intricsic[1][1]) / p[2] + self.intricsic[1][2]
I think this is the way to project the point to image, p[0] is u, and p[1] is v.
but I can't understand these line, can you explain? thanks!
inside_mask = (pi[0] >= 0) * (pi[1] >= 0)
* (pi[0] < self.imageDim[0]) * (pi[1] < self.imageDim[1])
occlusion_mask = np.abs(depth[pi[1][inside_mask], pi[0][inside_mask]]
- p[2][inside_mask]) <= self.voxel_size
inside_mask[inside_mask == True] = occlusion_mask
