TODO: Transform tiles to a stream of samples

[vsimko]

No description provided.

[aklink]

Prepared an example method "MultibandTile2LabelPoint" in UtilsSVM

• class id is missing (currently placeholder 0.0 = Unclassified)
• selection of "area of interest" per class missing
• not elegant way (stitch + parallelize afterwards = not memory efficient, workflow not parallelized)
• spatial information is lost: converting 2(+1) dimensional raster tile to 1(+1) dimensional point samples (note dimension+1: bands are converted to featurevector)

[aklink]

Since "Raster tile to pixels and back" is working, now the Spark Streaming / Kafka queue integration is missing.

[aklink]

TilePixelingExample can be extended by kafka producer in spark:

// Use UtilsKafka & UtilsSVM to create Kafka Stream from samples
    val (brokers, topic) = ("localhost:9092","neu")
    UtilsKafka.initKafka(topic)(brokers)
    val messages : Array[String] = UtilsSVM.toKafkaString(samples, UtilsSVM.Delimiter(";")).toLocalIterator.toArray
    val nmsg : Int = messages.length
    logger debug s"messages $nmsg"
    for ( imsg <- 0 until nmsg ){
      val str = messages(imsg)
      UtilsKafka.send(str)
    }

testet with: kafka_2.11-0.10.1.1$ bin/kafka-console-consumer.sh --bootstrap-server localhost:9092 --topic neu --from-beginning :

1.0;2109.0;1965.0;1857.0;2653.0;SpatialKey(69005,41753);36;144
1.0;1827.0;1733.0;1658.0;2315.0;SpatialKey(69005,41753);36;145
1.0;1727.0;1650.0;1587.0;2194.0;SpatialKey(69005,41753);36;146
1.0;1727.0;1650.0;1587.0;2194.0;SpatialKey(69005,41753);36;147
1.0;1727.0;1650.0;1587.0;2194.0;SpatialKey(69005,41753);36;148
1.0;1727.0;1650.0;1587.0;2194.0;SpatialKey(69005,41753);36;149
1.0;1727.0;1650.0;1587.0;2194.0;SpatialKey(69005,41753);36;150
1.0;1727.0;1650.0;1587.0;2194.0;SpatialKey(69005,41753);36;151
1.0;1727.0;1650.0;1587.0;2194.0;SpatialKey(69005,41753);36;152
1.0;1727.0;1650.0;1587.0;2194.0;SpatialKey(69005,41753);36;153
1.0;1727.0;1650.0;1587.0;2194.0;SpatialKey(69005,41753);36;154
1.0;1734.0;1647.0;1588.0;2189.0;SpatialKey(69005,41753);36;155
1.0;1740.0;1644.0;1590.0;2184.0;SpatialKey(69005,41753);36;156

but the "reverse" channel is still missing and metadata ist lost (e.g. zoomlevel, SRS, extent)