#transform #distance #euclidean #image #grid #binary #squared

distance-transform

An implementation of the Squared Euclidean distance transform

3 releases

Uses old Rust 2015

0.1.2 Feb 24, 2018
0.1.1 Feb 16, 2018
0.1.0 Feb 10, 2018

#11 in #euclidean


Used in chanvese

MIT/Apache

11KB
200 lines

This crate provides an implementation of a distance transform of binary grids using Squared Euclidean distances. It is a port of the C++ implementation of Distance Transforms of Sampled Functions by P. Felzenszwalb and D. Huttenlocher.

Examples

To use the functions inside module distance_transform::utils you need to compile this crate with the feature image-utils.

extern crate distance_transform;
extern crate image;

use distance_transform::*;
use distance_transform::utils;
use std::fs::File;

fn main() {
    // create a 128x128 binary image
    let imgwidth = 128usize;
    let imgheight = 128usize;
    let mut bimg = BoolGrid::new(imgwidth, imgheight);

    // draw a circle
    for (x, y, value) in bimg.iter_mut() {
        let pos = (x as isize - 64)*(x as isize - 64)
                + (y as isize - 64)*(y as isize - 64);
        *value = pos < 32*32 && pos > 31*31;
    }

    // do the distance transformation and
    // take the square root since squared distances are calculated
    let fmm = utils::sqrt_img(dt2d(&bimg));

    // scale values to range [0, 255] and save as image
    let fmm_img = utils::min_max_scaling(&fmm, &(0., 255.));
    let ref mut fmm_out = File::create("fmm_out.png").unwrap();
    utils::save_float_grid(&fmm_img, fmm_out, image::PNG).unwrap();
}

Dependencies

~0–325KB