wrong sum in Survey Area

[wolkstein]

hi,
i noticed a wrong result in calculated Area in m² if you create a Survey Polygon.
maybe this change in will fix getArea() in GeoTools. i am not sure if each polygon will calculated right. but some tests here produce good results. i use a known soccer field (FIFA-Norm105,00m x68,00m = 7.140,00 m²) to test the m² results.

new getArea() using simple Shoelace formula 2A = n SUM i=1(yi+ yi+1)(xi-xi+1). i use relative x and y values to reduce cpu load and prevent double overflow.

/**
     * Experimental Function, needs testing! Calculate the area of the polygon
     * 
     * @return area in m²
     */
    // TODO test and fix this function
    public static Area getArea(Polygon poly) {
        double sum = 0.0;
        int length = poly.getPoints().size();
        for (int i = 0; i < length - 1; i++) {
            // using relative x and y values reduce load
            // relative xi to x0, x0 = 0
            double x = latToMeters(poly.getPoints().get(i).getX()) - latToMeters(poly.getPoints().get(0).getX());
            // relative xi+1 to x0
            double xx = latToMeters(poly.getPoints().get(i + 1).getX()) - latToMeters(poly.getPoints().get(0).getX());
            // relative yi to y 0, y0 = 0
            double y = latToMeters(poly.getPoints().get(i).getY()) - latToMeters(poly.getPoints().get(0).getY());
            // relative yi+1 to y0
            double yy = latToMeters(poly.getPoints().get(i + 1).getY()) - latToMeters(poly.getPoints().get(0).getY());

            // Shoelace formula 2A = [n SUM i=1](yi+ yi+1)(xi-xi+1)
            // using variables here for better reading
            // no Math.abs here! trying "Math.abs((y + yy)*(x-xx)) / 2" produce random results in m² end sum.
            sum += (y + yy)*(x-xx) / 2;
        }
        // second A/2!? this is strange but works. we divide A / 2 in "sum += ..." two lines before
        return new Area(Math.abs(sum) / 2);
    }

/g
wolke

[arthurbenemann]

@wolkstein That's great. Do you want to submit this as a pull request via git/github? That way your name will be logged into the history of the project.

Also if you could send me a set of coordinates, and the area contained in them I can write a unit test to make sure that this function will always return the correct value. Especialy if someone changes it in the future.

[wolkstein]

hey,
i found a new problem with this code. if you draw the polygon counter-clockwise the result is wrong. :(. normally Shoelace formula produce positive values for each trapeze if the polygon points run cw and negative results if the points run in ccw direction. but after some more tests i noticed that points returned from poly produce sometimes positive and/or negative results in one for loop(one direction). maybe poly class returned points sorted. if i found more time, i will take a deeper look into this. at least calculating an polygon surface is a simple task with shoelace formula. so i do not understand why it makes so many problems here.

[arthurbenemann]

@wolkstein Give me a set of coordinates, and the area contained in those and I can write a test to help in the development.

[wolkstein]

ok, i will crate a test polygon with reference coordinates and area. i noticed that a soccer field is not so good because simple geometry. i think i will create something like a irregular pentagram.

[wolkstein]

hi,
here are an polygon. t looks like an paper-sailboat sillouette.

Point, Lat, Long
1, 51,0282781100882, 2,16705322265625
2, 51,0317878663996, 2,16293267905712
3, 51,0313285768174, 2,16915607452393
4, 51,0349709823585, 2,17203207314014
5, 51,0318104275059, 2,17568457126617
6, 51,0326774403918, 2,18168333172798
7, 51,028521872254, 2,18014307320118

Hopefully i do a right calculation the surface area is 502915m²
defined as polygon point 8 = point 1

[arthurbenemann]

@wolkstein I added the test case in #770. You can run if you checkout the https://github.com/arthurbenemann/droidplanner/tree/issue764 branch, and in eclipse select the CORE project and click "Run as Junit test". Here is the test file if you want to check:
https://github.com/arthurbenemann/droidplanner/blob/issue764/Core/test/org/droidplanner/core/helpers/geoTools/GeoToolsTest.java

With the code that you provided the result is:
junit.framework.AssertionFailedError: expected:<502915.0> but was:<389665.7608050064>

With the old code (current master):
junit.framework.AssertionFailedError: expected:<502915.0> but was:<799101.5992431641>