python 笔记：PyTrack（将GPS数据和OpenStreetMap数据进行整合）【官网例子解读】

论文笔记：PyTrack: A Map-Matching-Based Python Toolbox for Vehicle Trajectory Reconstruction_UQI-LIUWJ的博客-CSDN博客4

0 包的安装

官网的两种方式我都试过，装是能装成功，但是python import PyTrack包的时候还是显示找不到Pytrack

# conda
conda install pytrack # or PyPI
pip install PyTrack-lib

于是使用这种安装方式，成功了：

pip 另一种安装方法：python setup.py install_python setup.py bdist_wheel did not run successful_UQI-LIUWJ的博客-CSDN博客

1 GPS数据导入

import numpy as np
import pandas as pddf=pd.DataFrame([{'datetime':'23-04-21 16:46:19:583000','lat':43.759650,'lon':11.291561},{'datetime':'23-04-21 16:46:36:570000','lat':43.759645,'lon':11.291544},{'datetime':'23-04-21 16:46:52:647000','lat':43.759671,'lon':11.291162},{'datetime':'23-04-21 16:47:37:568000','lat':43.759677,'lon':11.291148},{'datetime':'23-04-21 16:47:49:639000','lat':43.759691,'lon':11.290932},{'datetime':'23-04-21 17:12:37:573000','lat':43.779596,'lon':11.254733},{'datetime':'23-04-21 17:12:51:592000','lat':43.779583,'lon':11.254295},{'datetime':'23-04-21 17:13:05:572000','lat':43.779206,'lon':11.253978},{'datetime':'23-04-21 17:13:20:592000','lat':43.779205,'lon':11.253974},{'datetime':'23-04-21 17:13:36:590000','lat':43.778464,'lon':11.253364}])
df

 2 GPS数据处理

latitude = df["lat"].to_list()
longitude = df["lon"].to_list()points = [(lat, lon) for lat, lon in zip(latitude, longitude)]
points
'''
[(43.75965, 11.291561),(43.759645, 11.291544),(43.759671, 11.291162),(43.759677, 11.291148),(43.759691, 11.290932),(43.779596, 11.254733),(43.779583, 11.254295),(43.779206, 11.253978),(43.779205, 11.253974),(43.778464, 11.253364)]
'''north, east = np.max(np.array(points),axis=0)
south, west = np.min(np.array(points),axis=0)
north,south,east,west
#(43.779596, 43.759645, 11.291561, 11.253364)

3 获得路网Graph

from pytrack.graph import graph, distance
from pytrack.analytics import visualizationbbox=distance.enlarge_bbox(north, south, west, east, 500)
G = graph.graph_from_bbox(*distance.enlarge_bbox(north, south, west, east, 500), simplify=True, network_type='drive')
#Downloaded 1,767.31kB

3.1 显示提取的图

map=visualization.Map(location=(np.mean(latitude),np.mean(longitude)))
map.add_graph(G,plot_nodes=True)
map

 4  提取候选点

• get_candidates方法返回两个输出
  • graph的插值版本G_interp
  • 候选点的字典candidates
• 下面0，1 ，7，8都是一个candidate；2，3，4，5，6是4个；9是3个

from pytrack.matching import candidate, mpmatching_utils, mpmatching# Extract candidates
G_interp, candidates = candidate.get_candidates(G, points, interp_dist=5, closest=True, radius=30)candidates
'''
{0: {'observation': (43.75965, 11.291561),'osmid': [135115743],'edge_osmid': [590848305],'candidates': [(43.75963763247117, 11.291543159878621)],'candidate_type': array([False]),'dists': [1.9859410359888336]},1: {'observation': (43.759645, 11.291544),'osmid': [135115743],'edge_osmid': [590848305],'candidates': [(43.75963763247117, 11.291543159878621)],'candidate_type': array([False]),'dists': [0.8220066596419578]},2: {'observation': (43.759671, 11.291162),'osmid': [1366820392, 831342769, 831342769, 245816382],'edge_osmid': [158183875, 160047415, 202760017, 590848305],'candidates': [(43.759700520685946, 11.290803894486569),(43.7596557, 11.290801),(43.7596557, 11.290801),(43.75964691271334, 11.29117208534078)],'candidate_type': array([False, False, False, False]),'dists': [28.94629461607051,29.041910651129065,29.041910651129065,2.798176615616366]},3: {'observation': (43.759677, 11.291148),'osmid': [1366820392, 831342769, 831342769, 245816382],'edge_osmid': [158183875, 160047415, 202760017, 590848305],'candidates': [(43.759700520685946, 11.290803894486569),(43.7596557, 11.290801),(43.7596557, 11.290801),(43.75964691271334, 11.29117208534078)],'candidate_type': array([False, False, False, False]),'dists': [27.7587065301145,27.968156168366924,27.968156168366924,3.864490374081274]},4: {'observation': (43.759691, 11.290932),'osmid': [1366820392, 831342769, 831342769, 1751266336],'edge_osmid': [158183875, 160047415, 202760017, 590848305],'candidates': [(43.759700520685946, 11.290803894486569),(43.7596557, 11.290801),(43.7596557, 11.290801),(43.75965309954146, 11.290924702315552)],'candidate_type': array([False, False, False, False]),'dists': [10.342517462643924,11.229036268613909,11.229036268613909,4.254901893095896]},5: {'observation': (43.779596, 11.254733),'osmid': [1082477038, 1196043488, 699408714, 1082477038],'edge_osmid': [23338921, 24695287, 24933636, 203924197],'candidates': [(43.7796997, 11.2544238),(43.77956558718794, 11.254738857110752),(43.779738600417716, 11.254457384444233),(43.7796997, 11.2544238)],'candidate_type': array([False, False, False, False]),'dists': [27.371092347460355,3.4142911971419823,27.222205203514605,27.371092347460355]},6: {'observation': (43.779583, 11.254295),'osmid': [1082477038, 1082477038, 1082477038, 432026454],'edge_osmid': [23338921, 24695287, 24933636, 203924197],'candidates': [(43.7796997, 11.2544238),(43.7796997, 11.2544238),(43.7796997, 11.2544238),(43.77958438083008, 11.254328000262998)],'candidate_type': array([False, False, False, False]),'dists': [16.59261899959034,16.59261899959034,16.59261899959034,2.6538256098028095]},7: {'observation': (43.779206, 11.253978),'osmid': [1531471656],'edge_osmid': [203924197],'candidates': [(43.77919997631304, 11.254008676575015)],'candidate_type': array([False]),'dists': [2.552298511854619]},8: {'observation': (43.779205, 11.253974),'osmid': [1531471656],'edge_osmid': [203924197],'candidates': [(43.77919997631304, 11.254008676575015)],'candidate_type': array([False]),'dists': [2.8394681876277263]},9: {'observation': (43.778464, 11.253364),'osmid': [568144517, 132324140, 132324140],'edge_osmid': [27317795, 491331618, 915510037],'candidates': [(43.77843868125, 11.25338133125),(43.7782499, 11.2532268),(43.7782499, 11.2532268)],'candidate_type': array([False, False, False]),'dists': [3.14040435143449, 26.231663705314208, 26.231663705314208]}}
'''

4.1 可视化candidate

maps.add_graph(G, plot_nodes=True)
maps.draw_candidates(candidates, 30)
maps

 

 5 创建Trellis图

根据上面的candidates创建相应的trellis

trellis = mpmatching_utils.create_trellis(candidates)

5.1 绘制Trellis图

trellis_draw = visualization.draw_trellis(trellis, figsize=(3, 7))
trellis_draw

每个点的candidate个数和前面candidates中每个点的candidate数目一致 

 6 进行地图匹配

# Perform the map-matching process
path_prob, predecessor = mpmatching.viterbi_search(G_interp, trellis, "start", "target")

6.1 绘制地图匹配图

# Plot map-matching results
maps.draw_path(G_interp, trellis, predecessor)
maps