我在android中有一个应用程序,在地图上加载了5000多个地方。我当时在考虑将NonHierarchicalViewBasedAlgorithm与可扩展NonHierarchicalDistanceBasedAlgorithm的聚类一起使用,但问题是客户只想先加载优先级较高的项目(不聚类-因此我们应该在启动时显示所有约500个位置),同时进行足够的缩放(例如zoomlevel) > 10),我们应该在地图上添加所有优先级的所有可见项。 因此,我自定义了NonHierarchicalDistanceBasedAlgorithm,我目前不使用它来对项目进行聚类(之所以这样做,是因为我需要四叉树逻辑来仅在地图中添加可见位置) 这是我的自定义内容:
@Override
public Set<? extends Cluster<T>> getClusters(double zoom) {
final int discreteZoom = (int) zoom;
final double zoomSpecificSpan = mMaxDistance / Math.pow(2, discreteZoom) / 256;
final Set<QuadItem<T>> visitedCandidates = new HashSet<QuadItem<T>>();
final Set<Cluster<T>> results = new HashSet<Cluster<T>>();
final Map<QuadItem<T>, Double> distanceToCluster = new HashMap<QuadItem<T>, Double>();
final Map<QuadItem<T>, StaticCluster<T>> itemToCluster = new HashMap<QuadItem<T>, StaticCluster<T>>();
synchronized (mQuadTree) {
for (QuadItem<T> candidate : getClusteringItems(mQuadTree, discreteZoom)) {
if (visitedCandidates.contains(candidate)) {
// Candidate is already part of another cluster.
continue;
}
Bounds searchBounds = createBoundsFromSpan(candidate.getPoint(), zoomSpecificSpan);
Collection<QuadItem<T>> clusterItems;
clusterItems = mQuadTree.search(searchBounds);
//if (clusterItems.size() == 1 ) { //I commented out this part
// Only the current marker is in range. Just add the single item to the results.
results.add(candidate);
visitedCandidates.add(candidate);
distanceToCluster.put(candidate, 0d);
continue;
/*Also commented out this so as to not cluster}
StaticCluster<T> cluster = new StaticCluster<T>(candidate.mClusterItem.getPosition());
results.add(cluster);
for (QuadItem<T> clusterItem : clusterItems) {
Double existingDistance = distanceToCluster.get(clusterItem);
double distance = distanceSquared(clusterItem.getPoint(), candidate.getPoint());
if (existingDistance != null) {
// Item already belongs to another cluster. Check if it's closer to this cluster.
if (existingDistance < distance) {
continue;
}
// Move item to the closer cluster.
itemToCluster.get(clusterItem).remove(clusterItem.mClusterItem);
}
distanceToCluster.put(clusterItem, distance);
cluster.add(clusterItem.mClusterItem);
itemToCluster.put(clusterItem, cluster);
}
visitedCandidates.addAll(clusterItems); */
}
}
return results;
}
我知道我的实现不是最好的,也不是正确的(我可以从地图的速度看到它,这就是为什么我要提出任何想法!