使用OsmSharp处理OpenStreetMap(OSM)数据

欢迎访问我的博客GISer空间。

1. OpenStreetMap数据模型

        OSM数据有三个基本对象：Node，Way，Relation。更多信息请查看OSM-wiki。

2. OsmSharp

        OsmSharp可以在直接.NET中使用OSM数据，其主要功能有：

• Read/Write OSM-XML.
• Read/Write OSM-PBF.
• Streamed architecture, minimal memory footprint.
• Convert a stream of native OSM objects to ‘complete’ OSM objects: Ways with all their actual nodes, Relations with all members instantiated.
• Convert OSM objects to geometries.

3. 相关项目

        OsmSharp做过与地图相关的很多事情：路径规划、数据处理、渲染矢量数据。下面是OsmSharp与.NET平台的协作项目：

        NetTopologySuite(NTS)：一个地理库，可以将其与OsmSharp.Geo一起使用，将OSM数据转换为shapefile或过滤一些数据将其转换为GeoJSON。

        ltinero：.NET的路径规划项目。OsmSharp.Routing作为OsmSharp的一部分，现更名为ltinero。

        Mapsui：Mapsui是WPF，Xamarin.Android，Xamarin.iOS和UWP应用程序的.NET Map组件。

4. 流模型(Streaming Model)

        OsmSharp使用流模型处理OSM数据。所有流模型都实现了泛型IEnumerable接口，这就意味着可以使用LINQ查询处理。

• OsmStreamSource：
  • XmlOsmStreamSource：读OSM-XML文件
  • PBFOsmStreamSource：读OSM-PBF文件
• OsmStreamTarget：
  • XmlOsmStreamTarget：写OSM-XML文件
  • PBFOsmStreamTarget：写OSM-PBF文件
• OsmStreamFilter：
  • OsmStreamFilterDelegate
  • OsmStreamFilterMerge
  • OsmStreamFilterNode
  • OsmStreamFilterProgress

4.1 读OSM-PBF文件

luxembourg-latest.osm.pbf

// 读取"luxembourg-latest.osm.pbf"的node、way、relation个数

// 创建StreamSource
var source = new PBFOsmStreamSource(File.OpenRead("luxembourg-latest.osm.pbf"));
int nodes = 0, ways = 0, relations = 0;
foreach (var osmGeo in source)
{
	if (osmGeo.Type == OsmGeoType.Node)
	{
		nodes++;
	}
	if (osmGeo.Type == OsmGeoType.Way)
	{
		ways++;
	}
	if (osmGeo.Type == OsmGeoType.Relation)
	{
		relations++;
	}
}
Console.WriteLine("There are {0} nodes, {1} ways, {2} relations.", nodes, ways, relations);
// 输出：There are 1721051 nodes, 223718 ways, 2511 relations.

4.2 LINQ查询过滤数据

// 创建StreamSource
var source = new PBFOsmStreamSource(File.OpenRead("luxembourg-latest.osm.pbf"));
int nodes = 0, ways = 0, relations = 0;
// 使用LINQ查询用户名为"Stilmant Michael"的数据
var filtered = from osmGeo in source where osmGeo.UserName == "Stilmant Michael" select osmGeo;
foreach (var osmGeo in filtered)
{
	if (osmGeo.Type == OsmGeoType.Node)
	{
		nodes++;
	}
	if (osmGeo.Type == OsmGeoType.Way)
	{
		ways++;
	}
	if (osmGeo.Type == OsmGeoType.Relation)
	{
		relations++;
	}
}
Console.WriteLine("There are {0} nodes, {1} ways, {2} relations are edited by Stilmant Michael.", nodes, ways, relations);
// 输出：There are 762 nodes, 77 ways, 0 relations are edited by Stilmant Michael.

4.3 写入OSM-XML文件 

// 创建StreamSource
var source = new PBFOsmStreamSource(File.OpenRead("luxembourg-latest.osm.pbf"));
// 过滤
var filterd = from osmGeo in source where osmGeo.UserName == "Stilmant Michael" select osmGeo;
// 创建StreamTarget
var target = new XmlOsmStreamTarget(File.Open("filtered.osm", FileMode.Create));
// 写入
target.RegisterSource(filterd);
target.Pull();

 打开"filtered.osm"文件可查看内容，如下：

5. 常用数据处理

5.1 裁剪

        以下两种过滤方法可实现裁剪。

• FilterBox(float left, float top, float right, float bottom，bool completeWays)
• FilterSpatial(IPolygon polygon, bool completeWays)

        其中，FilterBox()以边框过滤，需设置左、上、右、下的经纬度。FilterSpatial()以ploygon裁剪，保留polygon以内的所有对象，该方法需要引用OsmSharp.Geo。completeWays默认为false，true和false的区别如下图。（红色为设置为true的结果，绿色为设置为false的结果）

示例：

using (var fileStreamSource = File.OpenRead("luxembourg-latest.osm.pbf"))
{
	using (var fileStreamTarget = File.Open("clip.osm", FileMode.Create))
	{
		// 创建StreamSource
		var source = new PBFOsmStreamSource(fileStreamSource);
		// 创建StreamTarget
		var target = new XmlOsmStreamTarget(fileStreamTarget);
		// 1.以边框过滤
		//var filtered = source.FilterBox(6.238002777099609f, 49.72076145492323f, 6.272850036621093f, 49.69928180928878f);

		// 2.以polygon过滤
		var polygon = GetPolygonFromGeoJson("polygon.geojson");
		var filtered = source.FilterSpatial(polygon, true);
		target.RegisterSource(filtered);
		target.Pull();
	}
}

GetPolygonFromGeoJson(string fileName)方法如下，需添加以下引用。

using GeoAPI.Geometries;
using NetTopologySuite.Features;
using Newtonsoft.Json;

/// 

/// 从GeoJson中获取Polygon对象
/// 
/// 文件名
/// 
private static IPolygon GetPolygonFromGeoJson(string fileName)
{
	using (var stream = new StreamReader(fileName))
	{
		var jsonSerializer = new NetTopologySuite.IO.GeoJsonSerializer();
		var features = jsonSerializer.Deserialize(new JsonTextReader(stream));
		return features.Features[0].Geometry as IPolygon;
	}
}

6. 数据库

6.1 SQLServer

        OsmSharp-GitHub上有两种方式从/向SQL Server数据库读取/写入OSM数据。

• OsmSharp.Db.SQLServer.dll
• OsmSharp.Data.SQLServer.dll

6.1.1 OsmSharp.Db.SQLServer.dll

       使用sqlserver-dataprovider软件包，以SQL Server作为OpenStreetMap数据库。从/向SQL Server数据库读取/写入OSM数据。需添加引用"OsmSharp.Db.SQLServer.dll"，最新的GitHub上找不到这个文件，我上传到自己的GitHub了，点击即可下载。

①向SQL Server数据库写入OSM数据

        实时数据(Snapshot*)和历史数据(History*)的插入方法一致，只是数据表的结构不一样。具体的数据表结构请查看GitHub中的SQL语句。

历史数据(History*)的导入，会因为id重复，而导致导入数据失败！！！

// 连接字符串
string conStr = "Server=*;Database=*;User Id=*;Password=*;MultipleActiveResultSets=true;";
// 启用日志记录
OsmSharp.Logging.Logger.LogAction = (origin, level, message, parameters) =>
{
	Console.WriteLine("[{0}-{3}]:{1} - {2}", origin, level, message, DateTime.Now.ToString());
};
// 连接数据库并插入数据
using (var connection = new SqlConnection(conStr))
{
	connection.Open();
	// 创建数据表
	Tools.SnapshotDbCreateAndDetect(connection);
	// 删除数据表内容
	Tools.SnapshotDbDeleteAll(connection);
	using (var stream = File.OpenRead("luxembourg-latest.osm.pbf"))
	{
		var source = new PBFOsmStreamSource(stream);
		var target = new SnapshotDbStreamTarget(conStr, true);
		// 插入数据
		target.RegisterSource(source);
		target.Pull();
	}
}

输出的日志：

[Schema.Tools-2017/5/12 13:32:47]:information - Delete all data in snapshot database schema...
[SnapshotDbStreamTarget-2017/5/12 13:32:53]:information - Inserting 1000000 records into node.
[SnapshotDbStreamTarget-2017/5/12 13:33:14]:information - Inserting 80777 records into node_tags.
[SnapshotDbStreamTarget-2017/5/12 13:33:22]:information - Inserting 100000 records into way.
[SnapshotDbStreamTarget-2017/5/12 13:33:23]:information - Inserting 354601 records into way_tags.
[SnapshotDbStreamTarget-2017/5/12 13:33:28]:information - Inserting 1122157 records into way_nodes.
[SnapshotDbStreamTarget-2017/5/12 13:33:43]:information - Inserting 100000 records into way.
[SnapshotDbStreamTarget-2017/5/12 13:33:44]:information - Inserting 256525 records into way_tags.
[SnapshotDbStreamTarget-2017/5/12 13:33:47]:information - Inserting 888866 records into way_nodes.
[SnapshotDbStreamTarget-2017/5/12 13:33:59]:information - Flushing remaining data...
[SnapshotDbStreamTarget-2017/5/12 13:33:59]:information - Inserting 721051 records into node.
[SnapshotDbStreamTarget-2017/5/12 13:34:15]:information - Inserting 55585 records into node_tags.
[SnapshotDbStreamTarget-2017/5/12 13:34:16]:information - Inserting 23718 records into way.
[SnapshotDbStreamTarget-2017/5/12 13:34:16]:information - Inserting 52142 records into way_tags.
[SnapshotDbStreamTarget-2017/5/12 13:34:16]:information - Inserting 282298 records into way_nodes.
[SnapshotDbStreamTarget-2017/5/12 13:34:19]:information - Inserting 2511 records into relation.
[SnapshotDbStreamTarget-2017/5/12 13:34:20]:information - Inserting 14759 records into relation_tags.
[SnapshotDbStreamTarget-2017/5/12 13:34:20]:information - Inserting 146095 records into relation_members.
[SnapshotDbStreamTarget-2017/5/12 13:34:22]:information - Database connection closed.

实时数据的表结构和插入后的数据如下图：

②从SQL Server数据库中读取OSM数据

// 连接字符串
string conStr = "Server=*;Database=*;User Id=*;Password=*;MultipleActiveResultSets=true;";
using (var connection = new SqlConnection(conStr))
{
	connection.Open();
	// 从数据库中读取OSM数据
	var source = new SnapshotDbStreamSource(conStr);
	int nodes = 0, ways = 0, relations = 0;
	foreach (var osmGeo in source)
	{
		if (osmGeo.Type == OsmGeoType.Node)
		{
			nodes++;
		}
		if (osmGeo.Type == OsmGeoType.Way)
		{
			ways++;
		}
		if (osmGeo.Type == OsmGeoType.Relation)
		{
			relations++;
		}
	}
	Console.WriteLine("There are {0} nodes, {1} ways, {2} relations.", nodes, ways, relations);
}
// 输出：There are 1721051 nodes, 223718 ways, 2511 relations.

6.1.2 OsmSharp.Data.SQLServer.dll

        使用data-providers软件包，可以下载“OsmSharp.Data.SQLServer.dll”文件，添加引用，也可以用NuGet搜索“OsmSharp.SQLServer”安装。

最好使用NuGet安装，这样会自动添加对应版本的OsmSharp库，否则可能出现某种问题。以下是本文所对应的版本号。

①SQL Server数据库写入OSM数据

        OsmSharp.Data.SQLServer.dll并没有对id作限制，因此有重复id的历史数据也可以直接写入数据库中。

history.pbf

string conStr = "Server=*;Database=*;User Id=*;Password=*;MultipleActiveResultSets=true;";
using (var connection = new SqlConnection(conStr))
{
	connection.Open();
	// 删除原有数据表
	SQLServerSchemaTools.Remove(connection);
	// 创建数据表
	SQLServerSchemaTools.CreateAndDetect(connection);
	// 实时数据
	// using (Stream stream = File.OpenRead("luxembourg-latest.osm.pbf"))
	// 历史数据
	using (Stream stream = File.OpenRead("history.pbf"))
	{
		var source = new PBFOsmStreamSource(stream);
		var target = new SQLServerOsmStreamTarget(connection);
		// 插入数据
		target.RegisterSource(source);
		target.Pull();
	}
}

        下图是历史数据写入的结果，可以看到有重复的id。

②SQL Server数据库读取OSM数据

        OsmSharp.Data.SQLServer.dll好像并不能从SQL Server数据库中读取OSM数据。

7. 路径规划

7.1 Itinero

        ltinero：.NET的路径规划项目。

• Itinero.IO.OSM：使用OSM数据
• Itinero.IO.Shape：使用shapefile
• Itinero.IO.Geo：使用NTS

        关键类：RouterDb，Router，Profiles 和 RouterPoint

• RouterDb: Manages the data of one routing network. It holds all data in RAM or uses a memory mapping strategy to load data on demand. It holds all the network geometry, meta data and topology.
• Router: The main facade for all routing functionality available. It will decide the best algorithm to use based on a combination of what's requested and what data is available in the RouterDb.
• Profiles: Definitions of vehicle and their behaviour that can traverse the routing network.
• RouterPoint: A location on the routing network to use as a start or endpoint of a route. It's defined by an edge-id and an offset-value uniquely identifying it's location on the network.

7.2 示例

        下面通过一个简单的示例介绍如何使用OSM数据做路径规划。

• 通过NuGet安装Itinero和Itinero.IO.OSM，添加引用

using Itinero;
using Itinero.IO.Osm;
using Itinero.LocalGeo;
using Itinero.Osm.Vehicles;

• 加载OSM数据

// 加载OSM数据
var routerDb = new RouterDb();
using (var stream = File.OpenRead("beijing.osm.pbf"))
{
	routerDb.LoadOsmData(stream, Vehicle.Car);
}

• 创建Router

var router = new Router(routerDb);

• 计算

var route = router.Calculate(Vehicle.Car.Fastest(), new Coordinate(39.9225f, 116.3669f), new Coordinate(39.9066f, 116.4053f));

// Vehicle
// public static readonly Profiles.Vehicle Car;
// public static readonly Profiles.Vehicle Pedestrian;
// public static readonly Profiles.Vehicle Bicycle;
// public static readonly Profiles.Vehicle Moped;
// public static readonly Profiles.Vehicle MotorCycle;
// public static readonly Profiles.Vehicle SmallTruck;
// public static readonly Profiles.Vehicle BigTruck;
// public static readonly Profiles.Vehicle Bus;

• 结果 

// route.TotalDistance; // 距离
// route.TotalTime;     // 时间
// route.ToGeoJson();   // 转为GeoJson格式:string
// route.WriteGeoJson();// 写入GeoJson文件
// ......

        将结果写入GeoJson文件后，添加到geojson.io，计算出的路径如下图。

7.3 实际应用

        上述示例对于小区域、城市或国家而言，没什么问题，但是当加载大面积数据时，请使用以下方法。

1. 处理原始数据并写入磁盘。
2. 从磁盘加载预处理的数据并将其用于路径规划。

• 加载原始数据

var routerDb = new RouterDb();
using (var stream = File.OpenRead("beijing.osm.pbf"))
{
	routerDb.LoadOsmData(stream, Vehicle.Car);
}

• 将routerDb实例写入到磁盘： 

using (var stream = File.Create("osmfile.routing"))
{
	routerDb.Serialize(stream);
}

• 加载routerDb

using (var stream = File.OpenRead("osmfile.routing"))
{
	var routerDb = RouterDb.Deserialize(stream, RouterDbProfile.NoCache);
}

• 计算 

var router = new Router(routerDb);

var profile = Vehicle.Car.Fastest();
var routerPoint1 = router.TryResolve(profile, 39.9225f, 116.3669f);
if(routerPoint1.IsError)
{
    // do something or retry.
}
var routerPoint2 = router.TryResolve(profile, 39.9066f, 116.4053f);
if (routerPoint2.IsError)
{
    // do something or retry.
}
var route = router.TryCalculate(Vehicle.Car.Fastest(),
    routerPoint1.Value, routerPoint2.Value);
if(route.IsError)
{
    // do something or retry.
}

        更过关于路径规划的内容请参考Itinero-GitHub。

 

更多精彩请关注GISer空间

 

 

未完待续...

转载于:https://my.oschina.net/huanglii/blog/897528