为了计算两条路线的相似度：
- 需要将“线”填充为“多边形”，加入面积属性。
- 再计算两个“多边形”的交集的面积占比。
如何将路线填充？因为路线是由点的序列构成的，计划由每个点生成一个六边形，再将一系列的六边形取并集，最终连接成路线形状的多边形。
//以每个原始点为中心的六边形
func InsertTrackPoint(userId, routeId, polygon string) (trackId int64, err error) {
polygon = "POLYGON((" + polygon
polygon = polygon + "))"
clog.Debugf("polygon %s", polygon)
sql := fmt.Sprintf("INSERT INTO route_track_point(user_id,route_id,geom) "+
"VALUES ('%s','%s',ST_PolygonFromText('%s',4326)) RETURNING id;", userId, routeId, polygon)
clog.Infof("sql: %s", sql)
err = GHotTestPg.QueryRow(sql).Scan(&trackId)
if err != nil {
clog.Errorf("pg error: %v", err)
return
}
return
}
//将上述六边形取并集，形成路线的填充多边形
func CreateTrackPolygon(userId, routeId string) (err error) {
sql := fmt.Sprintf("INSERT INTO route_track_polygon(user_id,route_id,polygon) SELECT * FROM "+
"(SELECT '%s','%s',ST_Union(route_track_point.geom) AS geom FROM route_track_point WHERE route_id = '%s') sq;", userId, routeId, routeId)
clog.Infof("sql: %s", sql)
_, err = GHotTestPg.Exec(sql)
if err != nil {
clog.Errorf("pg error: %v", err)
}
return
}
//对点序列的处理
//以每个原始点为中心的六边形
var offsets [6][2]float64 = [6][2]float64{
[2]float64{0.0001, 0},
[2]float64{0.00005, -0.0000866},
[2]float64{-0.00005, -0.0000866},
[2]float64{-0.0001, 0},
[2]float64{-0.00005, 0.0000866},
[2]float64{0.00005, 0.0000866},
}
var line string
for _, point := range pointList {
var start, polygon string
lon := point.Longitude
lat := point.Latitude
line += fmt.Sprintf("%f %f,", lon, lat)
for i, offset := range offsets {
if i == 0 {
start = fmt.Sprintf("%f %f,", lon+offset[0], lat+offset[1])
}
polygon += fmt.Sprintf("%f %f,", lon+offset[0], lat+offset[1])
}
polygon += start
if polygon != "" {
polygon = polygon[0:len(polygon) - 1]
InsertTrackPoint(req.UserId, req.RouteId, polygon)
}
}
CreateTrackPolygon(req.UserId, req.RouteId)
模拟结果如图：
