Calculate the center point of multiple latitude/longitude coordinate pairs

Viewed 141210

Given a set of latitude and longitude points, how can I calculate the latitude and longitude of the center point of that set (aka a point that would center a view on all points)?

EDIT: Python solution I've used:

Convert lat/lon (must be in radians) to Cartesian coordinates for each location.
X = cos(lat) * cos(lon)
Y = cos(lat) * sin(lon)
Z = sin(lat)

Compute average x, y and z coordinates.
x = (x1 + x2 + ... + xn) / n
y = (y1 + y2 + ... + yn) / n
z = (z1 + z2 + ... + zn) / n

Convert average x, y, z coordinate to latitude and longitude.
Lon = atan2(y, x)
Hyp = sqrt(x * x + y * y)
Lat = atan2(z, hyp)
23 Answers

Java Version if anyone needs it. Constants defined static to not calculate them twice.

/**************************************************************************************************************
 *   Center of geometry defined by coordinates
 **************************************************************************************************************/
private static double pi = Math.PI / 180;
private static double xpi = 180 / Math.PI;

public static Coordinate center(Coordinate... arr) {
    if (arr.length == 1) {
        return arr[0];
    }
    double x = 0, y = 0, z = 0;

    for (Coordinate c : arr) {
        double latitude = c.lat() * pi, longitude = c.lon() * pi;
        double cl = Math.cos(latitude);//save it as we need it twice
        x += cl * Math.cos(longitude);
        y += cl * Math.sin(longitude);
        z += Math.sin(latitude);
    }

    int total = arr.length;

    x = x / total;
    y = y / total;
    z = z / total;

    double centralLongitude = Math.atan2(y, x);
    double centralSquareRoot = Math.sqrt(x * x + y * y);
    double centralLatitude = Math.atan2(z, centralSquareRoot);

    return new Coordinate(centralLatitude * xpi, centralLongitude * xpi);
}

Here is the Android version based on @Yodacheese's C# answer using Google Maps api:

public static LatLng GetCentralGeoCoordinate(List<LatLng> geoCoordinates) {        
    if (geoCoordinates.size() == 1)
    {
        return geoCoordinates.get(0);
    }

    double x = 0;
    double y = 0;
    double z = 0;

    for(LatLng geoCoordinate : geoCoordinates)
    {
        double  latitude = geoCoordinate.latitude * Math.PI / 180;
        double longitude = geoCoordinate.longitude * Math.PI / 180;

        x += Math.cos(latitude) * Math.cos(longitude);
        y += Math.cos(latitude) * Math.sin(longitude);
        z += Math.sin(latitude);
    }

    int total = geoCoordinates.size();

    x = x / total;
    y = y / total;
    z = z / total;

    double centralLongitude = Math.atan2(y, x);
    double centralSquareRoot = Math.sqrt(x * x + y * y);
    double centralLatitude = Math.atan2(z, centralSquareRoot);

    return new LatLng(centralLatitude * 180 / Math.PI, centralLongitude * 180 / Math.PI);

}

in app build.gradle add:

implementation 'com.google.android.gms:play-services-maps:17.0.0'

Here is the python Version for finding center point. The lat1 and lon1 are latitude and longitude lists. it will retuen the latitude and longitude of center point.

import numpy as np

def GetCenterFromDegrees(lat1,lon1):
    if (len(lat1) <= 0):
        return false;

    num_coords = len(lat1)
    X = 0.0
    Y = 0.0
    Z = 0.0

    for i in range (len(lat1)):
        lat = lat1[i] * np.pi / 180
        lon = lon1[i] * np.pi / 180

        a = np.cos(lat) * np.cos(lon)
        b = np.cos(lat) * np.sin(lon)
        c = np.sin(lat);

        X += a
        Y += b
        Z += c

    X /= num_coords
    Y /= num_coords
    Z /= num_coords

    lon = np.arctan2(Y, X)
    hyp = np.sqrt(X * X + Y * Y)
    lat = np.arctan2(Z, hyp)

    newX = (lat * 180 / np.pi)
    newY = (lon * 180 / np.pi)
    return newX, newY

Dart Implementation for Flutter to find Center point for Multiple Latitude, Longitude.

import math package

import 'dart:math' as math;

Latitude and Longitude List

List<LatLng> latLongList = [LatLng(12.9824, 80.0603),LatLng(13.0569,80.2425,)];

LatLng getCenterLatLong(List<LatLng> latLongList) {
    double pi = math.pi / 180;
    double xpi = 180 / math.pi;
    double x = 0, y = 0, z = 0;

    if(latLongList.length==1)
    {
        return latLongList[0];
    }
    for (int i = 0; i < latLongList.length; i++) {
      double latitude = latLongList[i].latitude * pi;
      double longitude = latLongList[i].longitude * pi;
      double c1 = math.cos(latitude);
      x = x + c1 * math.cos(longitude);
      y = y + c1 * math.sin(longitude);
      z = z + math.sin(latitude);
    }

    int total = latLongList.length;
    x = x / total;
    y = y / total;
    z = z / total;

    double centralLongitude = math.atan2(y, x);
    double centralSquareRoot = math.sqrt(x * x + y * y);
    double centralLatitude = math.atan2(z, centralSquareRoot);

    return LatLng(centralLatitude*xpi,centralLongitude*xpi);
}

I did this task in javascript like below

function GetCenterFromDegrees(data){
    // var data = [{lat:22.281610498720003,lng:70.77577162868579},{lat:22.28065743343672,lng:70.77624369747241},{lat:22.280860953131217,lng:70.77672113067706},{lat:22.281863655593973,lng:70.7762061465462}];
    var num_coords = data.length;
    var X = 0.0;
    var Y = 0.0;
    var Z = 0.0;

    for(i=0; i<num_coords; i++){
        var lat = data[i].lat * Math.PI / 180;
        var lon = data[i].lng * Math.PI / 180;
        var a = Math.cos(lat) * Math.cos(lon);
        var b = Math.cos(lat) * Math.sin(lon);
        var c = Math.sin(lat);

        X += a;
        Y += b;
        Z += c;
    }

    X /= num_coords;
    Y /= num_coords;
    Z /= num_coords;

    lon = Math.atan2(Y, X);
    var hyp = Math.sqrt(X * X + Y * Y);
    lat = Math.atan2(Z, hyp);

    var finalLat = lat * 180 / Math.PI;
    var finalLng =  lon * 180 / Math.PI; 

    var finalArray = Array();
    finalArray.push(finalLat);
    finalArray.push(finalLng);
    return finalArray;
}

Dart/Flutter Calculate the center point of multiple latitude/longitude coordinate pairs

Map<String, double> getLatLngCenter(List<List<double>> coords) {
    const LATIDX = 0;
    const LNGIDX = 1;
    double sumX = 0;
    double sumY = 0;
    double sumZ = 0;

    for (var i = 0; i < coords.length; i++) {
      var lat = VectorMath.radians(coords[i][LATIDX]);
      var lng = VectorMath.radians(coords[i][LNGIDX]);
      // sum of cartesian coordinates
      sumX += Math.cos(lat) * Math.cos(lng);
      sumY += Math.cos(lat) * Math.sin(lng);
      sumZ += Math.sin(lat);
    }

    var avgX = sumX / coords.length;
    var avgY = sumY / coords.length;
    var avgZ = sumZ / coords.length;

    // convert average x, y, z coordinate to latitude and longtitude
    var lng = Math.atan2(avgY, avgX);
    var hyp = Math.sqrt(avgX * avgX + avgY * avgY);
    var lat = Math.atan2(avgZ, hyp);

    return {
      "latitude": VectorMath.degrees(lat),
      "longitude": VectorMath.degrees(lng)
    };
  }

So many of these answers are just variations on an odd approach that doesn't find the true center of the bounding box that comprises all the points. Rather it finds the center of most points (a weighted center of sorts). If you want the true center of all points regardless of clustering and weights, you can get the bounding box and easily find the center of those 4 corners. If you aren't concerned about factoring in earth curvature, you can get away with something as simple as (C# code):

var lat = (coordinates.Min(x => x.lat) + coordinates.Max(x => x.lat))/2;
var lon = (coordinates.Min(x => x.lon) + coordinates.Max(x => x.lon))/2;
return new Tuple<double, double>(lat, lon);

As an appreciation for this thread, here is my little contribution with the implementation in Ruby, hoping that I will save someone a few minutes from their precious time:

def self.find_center(locations)

 number_of_locations = locations.length

 return locations.first if number_of_locations == 1

 x = y = z = 0.0
 locations.each do |station|
   latitude = station.latitude * Math::PI / 180
   longitude = station.longitude * Math::PI / 180

   x += Math.cos(latitude) * Math.cos(longitude)
   y += Math.cos(latitude) * Math.sin(longitude)
   z += Math.sin(latitude)
 end

 x = x/number_of_locations
 y = y/number_of_locations
 z = z/number_of_locations

 central_longitude =  Math.atan2(y, x)
 central_square_root = Math.sqrt(x * x + y * y)
 central_latitude = Math.atan2(z, central_square_root)

 [latitude: central_latitude * 180 / Math::PI, 
 longitude: central_longitude * 180 / Math::PI]
end

I used a formula that I got from www.geomidpoint.com and wrote the following C++ implementation. The array and geocoords are my own classes whose functionality should be self-explanatory.

/*
 * midpoints calculated using formula from www.geomidpoint.com
 */
   geocoords geocoords::calcmidpoint( array<geocoords>& points )
   {
      if( points.empty() ) return geocoords();

      float cart_x = 0,
            cart_y = 0,
            cart_z = 0;

      for( auto& point : points )
      {
         cart_x += cos( point.lat.rad() ) * cos( point.lon.rad() );
         cart_y += cos( point.lat.rad() ) * sin( point.lon.rad() );
         cart_z += sin( point.lat.rad() );
      }

      cart_x /= points.numelems();
      cart_y /= points.numelems();
      cart_z /= points.numelems();

      geocoords mean;

      mean.lat.rad( atan2( cart_z, sqrt( pow( cart_x, 2 ) + pow( cart_y, 2 ))));
      mean.lon.rad( atan2( cart_y, cart_x ));

      return mean;
   }

Scala version:

import scala.math._
    
case class Coordinate(latitude: Double, longitude: Double)
    
def center(coordinates: List[Coordinate]) = {
  val (a: Double, b: Double, c: Double) = coordinates.fold((0.0, 0.0, 0.0)) {
    case ((x: Double, y: Double, z: Double), coord: Coordinate) =>
      val latitude = coord.latitude * Pi / 180
      val longitude = coord.longitude * Pi / 180
      (x + cos(latitude) * cos(longitude), y + cos(latitude) * sin(longitude), z + sin(latitude))
  }
  val total = coordinates.length
  val (x: Double, y: Double, z: Double) = (a / total, b / total, c / total)
  val centralLongitude = atan2(y, x)
  val centralSquareRoot = sqrt(x * x + y * y)
  val centralLatitude = atan2(z, centralSquareRoot)
    
  Coordinate(centralLatitude * 180 / Pi, centralLongitude * 180 / Pi);
}

Related