geoskill-spatial-index-builder

空间索引构建 | Spatial Index Builder

Builds three kinds of spatial indexes and benchmarks their query performance against a batch of query windows; results from every index are strictly validated against brute-force scanning:

• R-tree: built on shapely's STRtree, performs exact intersection queries with predicate="intersects"; the most widely used spatial index in production environments.
• Quadtree: a self-implemented quadtree — recursively splits the bounding box into four quadrants, subdividing on capacity overflow; queries only traverse nodes intersecting the window, and features spanning child nodes are kept at the parent node to guarantee no misses.
• GeoHash: a self-implemented GeoHash encoder/decoder (base32) that builds an inverted index over all cells covered by each geometry's bounding box (avoiding misses caused by centroids falling outside the window); queries enumerate the cells covered by the window in grid alignment and then filter precisely.

The benchmark times each index and compares result consistency against brute-force scanning, reporting average latency, hit counts and speed-up ratio. --synthetic mode generates 300 random points.

Dependencies / 依赖

pip install numpy rasterio geopandas shapely fiona pyproj

Usage / 使用方法

Basic Usage

python geoskill-spatial-index-builder.py --bbox 116.0 39.0 117.0 40.0 --synthetic --output-dir ./out

Example 1 (synthetic data, 300 points, 15 queries, offline)

python geoskill-spatial-index-builder.py --bbox 116.0 39.0 117.0 40.0 --synthetic --features 300 --queries 15 --output-dir ./bench

Example 2: build and benchmark indexes on POI data

python geoskill-spatial-index-builder.py --input pois.shp --queries 30 --output-dir ./poi_bench

Example 3: high-precision GeoHash (precision 7)

python geoskill-spatial-index-builder.py --bbox 121.0 31.0 122.0 32.0 --synthetic --precision 7 --output-dir ./gh7 --quiet

Example 4: stress test with a large dataset

python geoskill-spatial-index-builder.py --bbox 116.0 39.0 117.0 40.0 --synthetic --features 2000 --queries 50 --output-dir ./stress

Example 5: low-precision GeoHash comparison

python geoskill-spatial-index-builder.py --bbox 116.0 39.0 117.0 40.0 --synthetic --precision 4 --features 500 --output-dir ./gh4

Output / 输出

Data Source / 数据源 / Source

• --input: local vector file
• --synthetic: locally generated random point set

Privacy / 隐私声明 / Privacy

• Runs offline by default; --synthetic mode requires no network at all.
• All processing is done locally; user data is never uploaded.

License / License

MIT

---

---

name: geoskill-spatial-index-builder description: '构建 R-tree / Quadtree / GeoHash 空间索引并统计查询性能，结果与暴力搜索对齐。Build R-tree / Quadtree / GeoHash spatial indexes, benchmark query performance and align results with brute-force search.'

空间索引构建 | Spatial Index Builder

构建三种空间索引并对一批查询窗口做性能基准测试，所有索引结果与暴力 扫描严格对齐：

• R-tree：基于 shapely STRtree，用 predicate="intersects" 做精确 相交查询，是生产环境最常用的空间索引。
• Quadtree：自实现四叉树——按外包矩形递归四分，容量溢出时细分，查询 只遍历与窗口相交的节点；跨子节点的要素留在父节点保证不漏检。
• GeoHash：自实现 GeoHash 编解码（base32），按几何外包矩形覆盖的所有 cell 建倒排表（避免质心在窗口外导致的漏检），查询时网格对齐枚举窗口 覆盖的 cell 再精确过滤。

基准测试对每个索引计时并与暴力扫描比对结果一致性，输出平均耗时、命中 数与加速比。--synthetic 模式生成 300 个随机点。

依赖

pip install numpy rasterio geopandas shapely fiona pyproj

使用方法

基本用法

python geoskill-spatial-index-builder.py --bbox 116.0 39.0 117.0 40.0 --synthetic --output-dir ./out

示例 1（合成数据，300 点 15 次查询，离线）

python geoskill-spatial-index-builder.py --bbox 116.0 39.0 117.0 40.0 --synthetic --features 300 --queries 15 --output-dir ./bench

示例 2：对 POI 数据建索引基准测试

python geoskill-spatial-index-builder.py --input pois.shp --queries 30 --output-dir ./poi_bench

示例 3：高精度 GeoHash（precision 7）

python geoskill-spatial-index-builder.py --bbox 121.0 31.0 122.0 32.0 --synthetic --precision 7 --output-dir ./gh7 --quiet

示例 4：大数据量压测

python geoskill-spatial-index-builder.py --bbox 116.0 39.0 117.0 40.0 --synthetic --features 2000 --queries 50 --output-dir ./stress

示例 5：低精度 GeoHash 对比

python geoskill-spatial-index-builder.py --bbox 116.0 39.0 117.0 40.0 --synthetic --precision 4 --features 500 --output-dir ./gh4

输出

数据源 / Source

• --input：本地矢量文件
• --synthetic：本地生成随机点集

隐私声明 / Privacy

• 默认离线运行，--synthetic 模式完全无网络。
• 所有处理在本地完成，不上传用户数据。

License

MIT