Merge pull request #235 from jpolchlo/feature/euclidean-distance-opti…

…mization Euclidean distance optimization
locationtech-labs · May 24, 2017 · fa11105 · fa11105
2 parents 7197511 + bb4f79b
commit fa11105
Show file tree

Hide file tree

Showing 3 changed files with 18 additions and 30 deletions.
diff --git a/docker/notebooks/Park citing.ipynb → docker/notebooks/Park siting.ipynb b/docker/notebooks/Park citing.ipynb → docker/notebooks/Park siting.ipynb
@@ -143,6 +143,7 @@
    },
    "outputs": [],
    "source": [
+    "# The following operation takes about 90 seconds on a reasonably capable 4-core laptop\n",
     "png_layer = PngRDD.makePyramid(weighted_layer, \"viridis\", end_zoom=8)"
    ]
   },
@@ -171,17 +172,9 @@
   {
    "cell_type": "code",
    "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "M.add_layer(VectorData(\"/tmp/bart.geojson\"), name=\"BART stops\", colors=[0xff0000])\n",
-    "M.add_layer(VectorData(\"/tmp/parks.geojson\"), name=\"Parks\", colors=[0xffff00])"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
+   "metadata": {
+    "collapsed": true
+   },
    "outputs": [],
    "source": [
     "M.layers"

diff --git a/geopyspark-backend/geotrellis/src/main/scala/geopyspark/geotrellis/TiledRasterRDD.scala b/geopyspark-backend/geotrellis/src/main/scala/geopyspark/geotrellis/TiledRasterRDD.scala
@@ -636,19 +636,19 @@ object SpatialTiledRasterRDD {
     SpatialTiledRasterRDD(None, MultibandTileLayerRDD(rdd.tileToLayout(metadata), metadata))
   }
 
-  def euclideanDistance(sc: SparkContext, geomWKT: String, srcCRSStr: String, requestedZoom: Int): TiledRasterRDD[SpatialKey]= {
+  def euclideanDistance(sc: SparkContext, geomWKT: String, geomCRSStr: String, cellTypeString: String, requestedZoom: Int): TiledRasterRDD[SpatialKey]= {
+    val cellType = CellType.fromName(cellTypeString)
     val geom = geotrellis.vector.io.wkt.WKT.read(geomWKT)
-    val srcCRS = TileRDD.getCRS(srcCRSStr).get
-    val LayoutLevel(z, ld) = ZoomedLayoutScheme(WebMercator).levelForZoom(requestedZoom)
+    val srcCRS = TileRDD.getCRS(geomCRSStr).get
+    val LayoutLevel(z, ld) = ZoomedLayoutScheme(srcCRS).levelForZoom(requestedZoom)
     val maptrans = ld.mapTransform
-    val reprojected = geom.reproject(srcCRS, WebMercator)
-    val GridBounds(cmin, rmin, cmax, rmax) = maptrans(reprojected.envelope)
+    val gb @ GridBounds(cmin, rmin, cmax, rmax) = maptrans(geom.envelope)
 
     val keys = for (r <- rmin to rmax; c <- cmin to cmax) yield SpatialKey(c, r)
 
     val pts = 
-      if (reprojected.isInstanceOf[MultiPoint]) {
-        reprojected.asInstanceOf[MultiPoint].points.map(_.jtsGeom.getCoordinate)
+      if (geom.isInstanceOf[MultiPoint]) {
+        geom.asInstanceOf[MultiPoint].points.map(_.jtsGeom.getCoordinate)
       } else {
         val coords = collection.mutable.ListBuffer.empty[Coordinate]
 
@@ -662,11 +662,10 @@ object SpatialTiledRasterRDD {
             coords += coord
           }
 
-          Rasterizer.foreachCellByGeometry(reprojected, re)(rasterizeToPoints)
-                                                           (sk, coords.toArray)
+          Rasterizer.foreachCellByGeometry(geom, re)(rasterizeToPoints)
         }
-        keys.foreach(createPoints)
 
+        keys.foreach(createPoints)
         coords.toArray
       }
 
@@ -675,19 +674,15 @@ object SpatialTiledRasterRDD {
     val mbtileRDD: RDD[(SpatialKey, MultibandTile)] = skRDD.mapPartitions({ skiter => skiter.map { sk =>
       val ex = maptrans(sk)
       val re = RasterExtent(ex, ld.tileCols, ld.tileRows)
-      val tile = DoubleArrayTile.empty(re.cols, re.rows)
+      val tile = ArrayTile.empty(cellType, re.cols, re.rows)
       val vd = new VoronoiDiagram(dt.value, ex)
       vd.voronoiCellsWithPoints.foreach(EuclideanDistanceTile.rasterizeDistanceCell(re, tile)(_))
       (sk, MultibandTile(tile))
     } }, preservesPartitioning=true)
 
-    val projectedRDD: RDD[(ProjectedExtent, MultibandTile)] = mbtileRDD.map{ case (sk, mbtile) => {
-      val ex = maptrans(sk)
-      val projEx = ProjectedExtent(ex, WebMercator)
-      (projEx, mbtile)
-    }}
+    val metadata = TileLayerMetadata(cellType, ld, maptrans(gb), srcCRS, KeyBounds(gb))
 
-    SpatialTiledRasterRDD(Some(z), MultibandTileLayerRDD(mbtileRDD, projectedRDD.collectMetadata[SpatialKey](ld)))
+    SpatialTiledRasterRDD(Some(z), MultibandTileLayerRDD(mbtileRDD, metadata))
   }
 
 }

diff --git a/geopyspark/geotrellis/rdd.py b/geopyspark/geotrellis/rdd.py
@@ -532,7 +532,7 @@ def rasterize(cls, geopysc, rdd_type, geometry, extent, crs, cols, rows,
         return cls(geopysc, rdd_type, srdd)
 
     @classmethod
-    def euclidean_distance(cls, geopysc, geometry, source_crs, zoom):
+    def euclidean_distance(cls, geopysc, geometry, source_crs, zoom, cellType='float64'):
         """Calculates the Euclidean distance of a Shapely geometry.
 
         Args:
@@ -553,7 +553,7 @@ def euclidean_distance(cls, geopysc, geometry, source_crs, zoom):
         if isinstance(source_crs, int):
             source_crs = str(source_crs)
 
-        srdd = geopysc._jvm.geopyspark.geotrellis.SpatialTiledRasterRDD.euclideanDistance(geopysc.sc, dumps(geometry), source_crs, zoom)
+        srdd = geopysc._jvm.geopyspark.geotrellis.SpatialTiledRasterRDD.euclideanDistance(geopysc.sc, dumps(geometry), source_crs, cellType, zoom)
         return cls(geopysc, SPATIAL, srdd)
 
     def to_numpy_rdd(self):