我试图创建一个快速的2D点内多边形算法,用于命中测试(例如多边形.contains(p:点))。对有效技术的建议将不胜感激。


当前回答

答案取决于你用的是简单多边形还是复杂多边形。简单多边形不能有任何线段交点。所以它们可以有洞,但线不能交叉。复杂区域可以有直线交点,所以它们可以有重叠的区域,或者只有一点相交的区域。

对于简单多边形,最好的算法是光线投射(交叉数)算法。对于复杂多边形,该算法不检测重叠区域内的点。所以对于复杂多边形你必须使用圈数算法。

下面是一篇用C实现这两种算法的优秀文章。我试过了,效果不错。

http://geomalgorithms.com/a03-_inclusion.html

其他回答

from typing import Iterable

def pnpoly(verts, x, y):
    #check if x and/or y is iterable
    xit, yit = isinstance(x, Iterable), isinstance(y, Iterable)
    #if not iterable, make an iterable of length 1
    X = x if xit else (x, )
    Y = y if yit else (y, )
    #store verts length as a range to juggle j
    r = range(len(verts))
    #final results if x or y is iterable
    results = []
    #traverse x and y coordinates
    for xp in X:
        for yp in Y:
            c = 0 #reset c at every new position
            for i in r:
                j = r[i-1] #set j to position before i
                #store a few arguments to shorten the if statement
                yneq       = (verts[i][1] > yp) != (verts[j][1] > yp)
                xofs, yofs = (verts[j][0] - verts[i][0]), (verts[j][1] - verts[i][1])
                #if we have crossed a line, increment c
                if (yneq and (xp < xofs * (yp - verts[i][1]) / yofs + verts[i][0])):
                    c += 1
            #if c is odd store the coordinates        
            if c%2:
                results.append((xp, yp))
    #return either coordinates or a bool, depending if x or y was an iterable
    return results if (xit or yit) else bool(c%2)

这个python版本是通用的。您可以为True/False结果输入单个x和单个y值,也可以使用x和y的范围来遍历整个点网格。如果使用范围,则返回所有True点的x/y对列表。vertices参数需要一个由x/y对组成的二维Iterable,例如:[(x1,y1), (x2,y2),…]

使用示例:

vertices = [(25,25), (75,25), (75,75), (25,75)]
pnpoly(vertices, 50, 50) #True
pnpoly(vertices, range(100), range(100)) #[(25,25), (25,26), (25,27), ...]

实际上,这些都可以。

pnpoly(vertices, 50, range(100)) #check 0 to 99 y at x of 50
pnpoly(vertices, range(100), 50) #check 0 to 99 x at y of 50

我已经做了nirg的c++代码的Python实现:

输入

Bounding_points:组成多边形的节点。 Bounding_box_positions:筛选的候选点。(在我从边界框创建的实现中。 (输入为元组列表,格式为:[(xcord, ycord),…])

返回

多边形内的所有点。

def polygon_ray_casting(self, bounding_points, bounding_box_positions):
    # Arrays containing the x- and y-coordinates of the polygon's vertices.
    vertx = [point[0] for point in bounding_points]
    verty = [point[1] for point in bounding_points]
    # Number of vertices in the polygon
    nvert = len(bounding_points)
    # Points that are inside
    points_inside = []

    # For every candidate position within the bounding box
    for idx, pos in enumerate(bounding_box_positions):
        testx, testy = (pos[0], pos[1])
        c = 0
        for i in range(0, nvert):
            j = i - 1 if i != 0 else nvert - 1
            if( ((verty[i] > testy ) != (verty[j] > testy))   and
                    (testx < (vertx[j] - vertx[i]) * (testy - verty[i]) / (verty[j] - verty[i]) + vertx[i]) ):
                c += 1
        # If odd, that means that we are inside the polygon
        if c % 2 == 1: 
            points_inside.append(pos)


    return points_inside

同样,这个想法也是从这里得来的

Like David Segonds' answer suggests I use an approach of angle summation derived from my concave polygon drawing algorithm. It relies of adding up the approximate angles of subtriangles around the point to obtain a weight. A weight around 1.0 means the point is inside the triangle, a weight around 0.0 means outside, a weight around -1.0 is what happens when inside the polygon but in reverse order (like with one of the halves of a bowtie-shaped tetragon) and a weight of NAN if exactly on an edge. The reason it's not slow is that angles don't need to be estimated accurately at all. Holes can be handled by treating them as separate polygons and subtracting the weights.

typedef struct { double x, y; } xy_t;

xy_t sub_xy(xy_t a, xy_t b)
{
    a.x -= b.x;
    a.y -= b.y;
    return a;
}

double calc_sharp_subtriangle_pixel_weight(xy_t p0, xy_t p1)
{
    xy_t rot, r0, r1;
    double weight;

    // Rotate points (unnormalised)
    rot = sub_xy(p1, p0);
    r0.x = rot.x*p0.y - rot.y*p0.x;
    r0.y = rot.x*p0.x + rot.y*p0.y;
    r1.y = rot.x*p1.x + rot.y*p1.y;

    // Calc weight
    weight = subtriangle_angle_approx(r1.y, r0.x) - subtriangle_angle_approx(r0.y, r0.x);

    return weight;
}

double calc_sharp_polygon_pixel_weight(xy_t p, xy_t *corner, int corner_count)
{
    int i;
    xy_t p0, p1;
    double weight = 0.;

    p0 = sub_xy(corner[corner_count-1], p);
    for (i=0; i < corner_count; i++)
    {
        // Transform corner coordinates
        p1 = sub_xy(corner[i], p);

        // Calculate weight for each subtriangle
        weight += calc_sharp_subtriangle_pixel_weight(p0, p1);
        p0 = p1;
    }

    return weight;
}

因此,对于多边形的每一段,都形成一个子三角形,并计算点,然后旋转每个子三角形以计算其近似角度并添加到权重。

调用subtriangle_angle_approx(y, x)可以替换为atan2(y, x) / (2.*pi),但是一个非常粗略的近似值就足够精确了:

double subtriangle_angle_approx(double y, double x)
{
    double angle, d;
    int obtuse;

    if (x == 0.)
        return NAN;

    obtuse = fabs(y) > fabs(x);
    if (obtuse)
        swap_double(&y, &x);

    // Core of the approximation, a very loosely approximate atan(y/x) / (2.*pi) over ]-1 , 1[
    d = y / x;
    angle = 0.13185 * d;

    if (obtuse)
        angle = sign(d)*0.25 - angle;

    return angle;
}

这似乎在R中工作(为丑陋道歉,希望看到更好的版本!)。

pnpoly <- function(nvert,vertx,verty,testx,testy){
          c <- FALSE
          j <- nvert 
          for (i in 1:nvert){
              if( ((verty[i]>testy) != (verty[j]>testy)) && 
   (testx < (vertx[j]-vertx[i])*(testy-verty[i])/(verty[j]-verty[i])+vertx[i]))
            {c <- !c}
             j <- i}
   return(c)}

David Segond's answer is pretty much the standard general answer, and Richard T's is the most common optimization, though therre are some others. Other strong optimizations are based on less general solutions. For example if you are going to check the same polygon with lots of points, triangulating the polygon can speed things up hugely as there are a number of very fast TIN searching algorithms. Another is if the polygon and points are on a limited plane at low resolution, say a screen display, you can paint the polygon onto a memory mapped display buffer in a given colour, and check the color of a given pixel to see if it lies in the polygons.

像许多优化一样,这些优化是基于特定情况而不是一般情况,并且基于摊销时间而不是单次使用产生效益。

在这个领域工作,我发现约瑟夫·奥鲁克斯的《计算几何》在C' ISBN 0-521-44034-3是一个很大的帮助。