fortran-simplify

Fortran module that reduces the over-sampled resolution of a polyline. This process can be useful when working with data that is noisy but usable with a significantly reduced resolution.

How To Use

The simplify module is all that is necessary to use the polyline simplification algorithms. The simplify module does depend on a smpl_precision module that at least needs to contain a real kind named dp. In this case, it uses the double precision real kind from the stdlib_kinds module found in stdlib. This can easily be modified to accept a real kind with another name in a different module by changing the line in simplify.f90 from

use smpl_precision

to

use real_kinds_module, only : dp => real_kind

All that is left is to add a use statement for the module in whichever program units you need the algorithms and to compile the module as normal.

To add fortran-simplify to your fpm project, simply add the following line to the fpm.toml file under the [dependencies] tag:

fortran-simplify.git = "https://github.com/jaiken17/fortran-simplify"

simplify_example Example Program

Included is a simple test program that runs the algorithms on a couple of curves. The two dimensional curve is also written to a file curve.data and the output from the perpendicular distance and Reumann-Witkam algorithms are written to perp_simple_curve.data and ruemann_witkam_simple_curve.data, respectively.

With the Fortran Package Manager (fpm) installed, compile and run the example program in a shell by excuting the following command in the top-level project directory:

fpm run --example

The data files are then created in the example directory and can then be plotted with gnuplot by running the plot bash script (within the example directory):

./plot

Function Signatures And Descriptions

For a better explanation of each algorithm see Polyline Simplification which also includes graphical representations to aid in understanding.

nth_point(curve, n)

Implementation of the nth point algorithm. Takes two intent(in) parameters:

• curve - a real(dp),dimension(:) or real(dp),dimension(:,:) array containing the points of the curve to be simplified. If real(dp), dimension(:,:), then curve(i,:) denotes each point of the curve (column major).
• n - a default integer that is interpreted as every nth element to be kept.

This function is of the same type as curve.

radial_distance(curve, tolerance)

Implementation of the radial distance algorithm. Takes two intent(in) parameters:

• curve - a real(dp),dimension(:) or real(dp),dimension(:,:) array containing the points of the curve to be simplified. If real(dp), dimension(:,:), then curve(i,:) denotes each point of the curve (column major).
• tolerance - a real(dp) value interpreted as the minimum distance required between any two consecutive points in the resulting curve.

This function is of the same type as curve.

perpendicular_distance(curve, tolerance, repeat)

Implementation of the perpendicular distance algorithm. Takes up to three intent(in) parameters:

• curve - a real(dp),dimension(:,:) array containing the points of the curve to be simplified. curve(i,:) denotes each point of the curve (column major).
• tolerance - a real(dp) value interpreted as the minimum distance a point can be from the line defined by the most recent confirmed key and the next point in the original curve.
• repeat - an optional, default integer that describes how many times to run the algorithm. This parameter is useful as the algorithm will at most remove 50% of the points from a given curve.

This function is of the same type as curve.

reumann_witkam(curve, tolerance)

Implementation of the Reumann-Witkam algorithm. Takes two parameters:

• curve - a real(dp),dimension(:,:) array containing the points of the curve to be simplified. curve(i,:) denotes each point of the curve (column major).
• tolerance - a real(dp) value interpreted as the minimum distance points must be from any line segment.

This function is of the same type as curve.