Research Article
Road-Type Classification with Deep AutoEncoder
Algorithm 2
Feature generation for each road segment.
| Require: and | | Output: Feature vector for each road segment | (1) | for each road segment do | (2) | Obtain the length of from . | (3) | Compute midpoint coordinates of . | (4) | Obtain geometry of . | (5) | if does not have a geometry then | (6) | Convert to line geometry. | (7) | Divide into 20 equally spaced distanced points . | (8) | for to 20 do | (9) | Subtract by midpoint coordinates . | (10) | end for | (11) | else | (12) | Divide geometry into 20 equally spaced distanced points . | (13) | for to 20 do | (14) | Subtract by midpoint coordinates . | (15) | end for | (16) | end if | (17) | Obtain one hot encoded vector of the speed limits with standard values. | (18) | Concatenate features generated from steps 2 to 17. | (19) | end for |
|