This should be simple but I am having a brain misfire. I want to read in a large ECW file and create a shape file which follows the raster data snuggly (not including the white, nodata areas). Obviously I need it to run as quickly as it can.
Not sure if it addresses your performance request, but maybe a PointCloudCombiner can be used to convert the raster to a pointcloud, a PointCloudFilter to remove the nodata values, and a PointCloudCoercer (Single Multipoint, no attributepreservation) to make it consumable for a final HullReplacer?
The RasterExtentsCoercer does exactly what you want (either MBR or Data Extents mode), but it may not be the most efficient transformer. Make sure the white pixels are set to NoData.
The RasterExtentsCoercer does exactly what you want (either MBR or Data Extents mode), but it may not be the most efficient transformer. Make sure the white pixels are set to NoData.
Not sure if it addresses your performance request, but maybe a PointCloudCombiner can be used to convert the raster to a pointcloud, a PointCloudFilter to remove the nodata values, and a PointCloudCoercer (Single Multipoint, no attributepreservation) to make it consumable for a final HullReplacer?
In either case you will get better performance if you turn your raster into a binary data/nodata. (Set everything valid value to 1, the white pixels to 0 and then set the NoDataValue to 0).
You could also look at the RasterToPolygonCoercer.
In either case you will get better performance if you turn your raster into a binary data/nodata. (Set everything valid value to 1, the white pixels to 0 and then set the NoDataValue to 0).
You could also look at the RasterToPolygonCoercer.
See https://knowledge.safe.com/articles/1226/using-conditions-with-the-rasterexpressionevaluato.html
Then the RasterBandNoDataSetter.