# Crossroads schematization
Generate a schematization of an intersection from OpenStreetMap data.
## Installation
With pip, use the following command line to install crschem:
* ```pip install crossroads-schematization```
Dependancies:
Crossroads schematization depends on ```crmodel```, that implies the following cascading dependancies:
* [crossroadsdescriber](https://github.com/jeremyk6/crmodel/)
* [crossroads-segmentation](https://github.com/jmtrivial/crossroads-segmentation)
* OSMnx
* NetworkX, geopandas
## Usage
If you installed crossroads-schematization using pip, a console script is now available using ```get_crossroad_schematization```.
This script is also available in the examples folder (```PYTHONPATH=$PWD examples/get-crossroad-schematization.py```). You will find a complete description of the parameters using ```--help```.
## Pipeline
First compute for each branch two long edges *S1* and *S2* corresponding to the sidewalks:
* for each edge part of a branch, identify the corresponding polyline (continue a couple of 10 meters outside of the crossing)
* fit a long edge *E* on this polyline, starting from the beginning of the polyline (e.g. 50 meters for example), and with a fixed length in the exterior direction)
* estimate the width of the way (using the number of lanes, the classification of the way)
* reconstruct a linear description of the sidewalk by shifting *E*
* if the branch is composed of more than one way, identify the two adjacent sidewalks and make them parallel
Each sidewalk is part of two branches (see crossroads-description by Jérémy Kalsron), thus is described by two long edges *[Sa_i, Sa_e]* and *[Sa_i, Sa_e]* (*e* for exterior, *i* for interior).
* we compute the intersection $m$ between *Sa* and *Sb*, and build a new representation with a polyline made of 3 points: *Sa_e*, *m*, *Sb_e*
* If a one of the original segments of the crossing intersects *Sa-m-Sb*, the sidewalk is described by a polyline made of 4 points: *Sa_e*, *Sa_i*, *Sb_i*, *Sb_e*
The inner part of the crossroad is computed by assembling all the sidewalks as a closed polyline (assembling each polyline at they extremity). From this region, we can apply a negative buffer to obtain the inner part of the road (with a white space between sidewalks and inner part of the road).
Each traffic island is available as a polygon from crossroads-description. We compute its compactness and size. A description is produced depending on these parameters:
* if the traffic island is large, the rendering is done as if it was bordered by sidewalks
* if the traffic island is small:
* if the shape is compact, we describe it by a disc
* if the shape is not compact, we describe it by two half discs connected by lines (a long shape)
Each pedestrian crossing is described by a dedicated pictogram, aligned with the corresponding edge from the initial data
