Adds a native k-means clustering algorithm.
Based on a port of PostGIS' ST_ClusterKMeans function, this
new algorithm adds a new cluster ID field to a set of input
features identify the feature's cluster based on the k-means
clustering approach. If non-point geometries are used as input,
the clustering is based off the centroid of the input geometries.
Converts a raster layer into a vector layer, with a polygon feature
corresponding to each pixel from the raster and a single field
containing the band value from the raster.
Sponsored by SMEC/SJ
Adds geometry methods to create variable width buffers, including
tapered buffers (with a specified start and end diameter) and
variable width buffers from line string m values.
Also adds processing algorithms which expose these methods
to processing.
This allows users to optionally set a sort order to use when
assigning values in the Add Incremental Field algorithm.
Previously values were always added using the original feature
order. With this change users can control the order in which
features are assigned values.
Finally starting a suite of unit tests for overlay algorithms:
- overlay1 - layers that cover various basic overlay situations
- overlay2 - layers where one input has self-intersecting polygons
- overlay3 - layers where intersections return different geometry types
Aside from the performance benefits, the Python version of this
algorithm occasionally fails on Travis with odd errors. Hopefully
by porting to c++ it will fix these, or at least give useful
debug information in the event of a fail.
Also add support for curved input geometries.
- Output interpolated points when travel cost falls mid-way along
an edge
- Output all intermediate reachable points also
- Make outputting upper/lower bound points optional, and non-default.
Now by default we just output all definitely reachable points and
the interpolated points along edges which correspond to the travel cost.
This allows the output to be used to correctly generate service areas
e.g. by concave/convex polygons and all reachable nodes will be
included in the area.
- Allow algorithm to optionally output a line layer (and make the
point layer optional too, and default to just the line layer output)
containing all reachable line segments (including interpolated
segments of lines when the travel cost sits midway along that
edge). This output is more easily understandably for users.
This algorithm swaps the X and Y coordinate values in input
geometries. It can be used to repair geometries which have
accidentally had their latitude and longitude values reversed.
This implements a new "import geotagged photos" algorithm
for processing. It allows selection of a folder which it
will scan for jpg files which have been geotagged and
creates a PointZ layer with the result, with attributes
for photo path, altitude, direction and timestamp.
Optionally the scan can be recursive and you can create
an optional table of photos which could not be read
or which were missing geotags.
The algorithm automatically sets the output table to
use an external resource widget to display the linked
photos in the attribute form.
[ALGCHANGE]
At this stage the raster processing tests just test that the
algorithm executes, and generates an output file - there's
no way to validate that the provided hashes are valid
outputs or not
Adds two new algorithms which expose QgsGeometry's methods
for segmentizing curved geometries.
"Segmentize by maximum distance":
The segmentization is performed by specifying the maximum
allowed offset distance between the original curve and the
segmentized representation.
"Segmentize by maximum angle":
The segmentization is performed by specifying the maximum
allowed radius angle between vertices on the straightened
geometry (e.g the angle of the arc created from the
original arc center to consective output vertices on the
linearized geometry).
