获取多边形经纬度顶点坐标 用于绘制Plotly Choropleth地图

需要从给定的经纬度散点列表中提取多边形顶点坐标，以便在Python中使用Plotly Choropleth绘制多边形。当前挑战是这些经纬度点分布在美国某城市范围内，较为分散。试过基于BFS算法的实现，但它仅适用于整数坐标且点需处于相邻网格；其他方案多采用OpenCV，需调整才能适配经纬度数据。考虑过正交复合凸包算法，但未找到对应的Python实现。

经纬度样本列表 #

# Sample list of latitudes and longitudes 
coords = [
    [-86.496774, 32.344437],
    [-86.717897, 32.402814],
    [-86.814912, 32.340803],
    [-86.890581, 32.502974],
    [-86.917595, 32.664169],
    [-86.71339, 32.661732],
    [-86.714219, 32.705694],
    [-86.413116, 32.707386],
    [-86.411172, 32.409937],
    [-86.496774, 32.344437],
    [-86.577799, 33.765316],
    [-86.759144, 33.840617],
    [-86.953664, 33.815297],
    [-86.954305, 33.844862],
    [-86.96296, 33.844865],
    [-86.963358, 33.858221],
    [-86.924387, 33.909222],
    [-86.793914, 33.952059],
    [-86.685365, 34.05914],
    [-86.692061, 34.092654],
    [-86.599632, 34.119914],
    [-86.514881, 34.25437],
    [-86.45302, 34.259317],
    [-86.303516, 34.099073],
    [-86.332723, 33.986109],
    [-86.370152, 33.93977],
    [-86.325622, 33.940147],
    [-86.377532, 33.861706],
    [-86.577528, 33.801977],
    [-86.577799, 33.765316]
]

解决方案与实现建议 #

1. 凸包算法（快速实现基础多边形） #

Scipy的ConvexHull支持浮点数坐标，能直接处理经纬度数据，适合提取点集的外边界。如果样本包含多个独立区域（比如你的数据里有两个闭合块），需要先聚类拆分点集再分别计算凸包。

代码示例（单区域凸包）

from scipy.spatial import ConvexHull
import numpy as np

# 转换为numpy数组
points = np.array(coords)
# 计算凸包
hull = ConvexHull(points)
# 提取凸包顶点（按逆时针顺序排列）
hull_vertices = points[hull.vertices].tolist()
# 闭合多边形（首尾点相同）
hull_vertices.append(hull_vertices[0])

print("凸包顶点坐标：")
print(hull_vertices)

处理多区域：先聚类再凸包

用DBSCAN根据地理距离聚类，拆分独立点集：

from sklearn.cluster import DBSCAN
from sklearn.metrics.pairwise import haversine_distances
import numpy as np

# 转换为弧度（haversine距离需要）
coords_rad = np.radians(points)
# 计算两两地理距离（单位：米）
distances = haversine_distances(coords_rad) * 6371000  # 地球半径6371km

# DBSCAN聚类：eps设为500米（可根据点密度调整），min_samples设为3
dbscan = DBSCAN(eps=500, min_samples=3, metric="precomputed")
clusters = dbscan.fit_predict(distances)

# 对每个簇计算凸包
polygons = []
for cluster_id in np.unique(clusters):
    if cluster_id == -1:
        continue  # 跳过噪声点
    cluster_points = points[clusters == cluster_id]
    hull = ConvexHull(cluster_points)
    cluster_polygon = cluster_points[hull.vertices].tolist()
    cluster_polygon.append(cluster_polygon[0])
    polygons.append(cluster_polygon)

print("多区域多边形顶点：")
for i, poly in enumerate(polygons):
    print(f"区域{i+1}：{poly}")

2. Alpha Shape算法（生成贴合的凹多边形） #

凸包只能生成凸边界，Alpha Shape可以生成更贴合实际的凹多边形，适合城市这类不规则区域。使用第三方库alphashape实现：

代码示例

import alphashape
from shapely.geometry import Polygon

# 生成alpha shape多边形，alpha值越小边界越贴合（需根据点密度调整）
alpha = 0.001
shape = alphashape.alphashape(coords, alpha)

# 提取顶点坐标（如果是MultiPolygon，需遍历每个子多边形）
if isinstance(shape, Polygon):
    polygon_vertices = list(shape.exterior.coords)
else:
    polygon_vertices = []
    for poly in shape.geoms:
        polygon_vertices.append(list(poly.exterior.coords))

print("Alpha Shape多边形顶点：")
print(polygon_vertices)

3. Plotly Choropleth适配 #

将提取的多边形顶点转换为Plotly要求的GeoJSON格式，即可用于Choropleth绘制：

import plotly.express as px
import json

# 构造GeoJSON
geojson = {
    "type": "FeatureCollection",
    "features": []
}

for i, poly in enumerate(polygons):
    feature = {
        "type": "Feature",
        "properties": {"id": i+1},
        "geometry": {
            "type": "Polygon",
            "coordinates": [poly]
        }
    }
    geojson["features"].append(feature)

# 绘制Choropleth
fig = px.choropleth_mapbox(
    geojson=geojson,
    locations=[f"{i+1}" for i in range(len(polygons))],
    featureidkey="properties.id",
    mapbox_style="carto-positron",
    center={"lat": 33.0, "lon": -86.6},
    zoom=8
)
fig.show()

内容的提问来源于stack exchange，提问作者prasanth_ntu