implement convert to/from euclidean/homogeneous + tests

.gitignore

@@ -1,8 +1,9 @@
 build/
 dist/
-torchvision.egg-info/
+torchgeometry.egg-info/
 */**/__pycache__
 */**/*.pyc
 */**/*~
 *~
+*.swp
 docs/build

test/test_functional.py

@@ -0,0 +1,24 @@
+import unittest
+
+import torch
+import torchgeometry as dgm
+
+
+class Tester(unittest.TestCase):
+
+    def test_convert_points_to_homogeneous(self):
+        points = torch.rand(1, 2, 3)
+
+        points_h = dgm.convert_points_to_homogeneous(points)
+        assert (points_h[..., -1] == torch.ones(1, 2, 1)).all()
+
+    def test_convert_points_from_homogeneous(self):
+        points_h = torch.rand(1, 2, 3)
+        points_h[..., -1] = 1.0
+
+        points = dgm.convert_points_from_homogeneous(points_h)
+        assert (points_h[..., :2] == points[..., :2]).all()
+
+
+if __name__ == '__main__':
+    unittest.main()

torchgeometry/__init__.py

@@ -1 +1,3 @@
-__version__ = '0.0.1'
+__version__ = '0.0.1'  # the current version of the lib
+
+from .functional import *

torchgeometry/functional.py

@@ -0,0 +1,41 @@
+import torch
+
+
+def convert_points_from_homogeneous(points, eps=1e-6):
+    """Converts points from homogeneous to Euclidean space.
+
+    Args:
+        points (Tensor): tensor of N-dimensional points of size (B, D, N).
+
+    Returns:
+        Tensor: tensor of N-1-dimensional points of size (B, D, N-1).
+    """
+    if not torch.is_tensor(points):
+        raise TypeError("Input type is not a torch.Tensor. Got {}"
+                        .format(type(points)))
+
+    if len(points.shape) != 3:
+        raise ValueError("Input size must be a three dimensional tensor. Got {}"
+                         .format(points.shape))
+
+    return points[..., :-1] / (points[..., -1:] + eps)
+
+
+def convert_points_to_homogeneous(points):
+    """Converts points from Euclidean to homogeneous space.
+
+    Args:
+        points (Tensor): tensor of N-dimensional points of size (B, D, N).
+
+    Returns:
+        Tensor: tensor of N+1-dimensional points of size (B, D, N+1).
+    """
+    if not torch.is_tensor(points):
+        raise TypeError("Input ype is not a torch.Tensor. Got {}"
+                        .format(type(points)))
+
+    if len(points.shape) != 3:
+        raise ValueError("Input size must be a three dimensional tensor. Got {}"
+                         .format(points.shape))
+
+    return torch.cat([points, torch.ones_like(points)[..., :1]], dim=-1)