size_divergence.c

Go to the documentation of this file.

/*
 * Copyright (C) 2015 Guido de Croon <guido.de.croon@gmail.com>
 *
 * From:
 * Characterization of Flow Field Divergence for Vertical Landing Control of MAVs
 * by H.W. Ho and G.C.H.E. de Croon (submitted)
 *
 * This file is part of Paparazzi.
 *
 * Paparazzi is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2, or (at your option)
 * any later version.
 *
 * Paparazzi is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with Paparazzi; see the file COPYING.  If not, see
 * <http://www.gnu.org/licenses/>.
 */

#include "size_divergence.h"
#include <stdlib.h>

#define NO_DIV 0.0

float get_size_divergence(struct flow_t *vectors, int count, int n_samples)
{
  float distance_1;
  float distance_2;
  float *divs;
  unsigned int sample;
  float dx;
  float dy;
  float mean_divergence;
  int n_elements;
  unsigned int i, j;

  // TODO: if count!/(2!*count-2!) lower than n_samples, do full sampling
  if (count < 2) {
    return NO_DIV;
  }

  if (n_samples == 0) {
    // divs will contain the individual divergence estimates:
    n_elements = (count * count - count) / 2;
    divs = (float *) malloc(sizeof(float) * n_elements);

    // go through all possible lines:
    sample = 0;
    for (i = 0; i < count; i++) {
      for (j = i + 1; j < count; j++) {
        // distance in previous image:
        dx = (float)vectors[i].pos.x - (float)vectors[j].pos.x;
        dy = (float)vectors[i].pos.y - (float)vectors[j].pos.y;
        distance_1 = sqrt(dx * dx + dy * dy);

        // distance in current image:
        dx = (float)vectors[i].pos.x + (float)vectors[i].flow_x - (float)vectors[j].pos.x - (float)vectors[j].flow_x;
        dy = (float)vectors[i].pos.y + (float)vectors[i].flow_y - (float)vectors[j].pos.y - (float)vectors[j].flow_y;
        distance_2 = sqrt(dx * dx + dy * dy);

        // calculate divergence for this sample:
        divs[sample] = (distance_2 - distance_1) / distance_1;
        sample++;
      }
    }

    // calculate the mean divergence:
    mean_divergence = get_mean(divs, n_elements);

    // free the memory of divs:
    free(divs);
  } else {
    // vector that will contain individual divergence estimates:
    divs = (float *) malloc(sizeof(float) * n_samples);

    // take random samples:
    for (sample = 0; sample < n_samples; sample++) {
      // take two random indices:
      i = rand() % count;
      j = rand() % count;
      // ensure it is not the same index:
      while (i == j) {
        j = rand() % count;
      }

      // distance in previous image:
      dx = (float)vectors[i].pos.x - (float)vectors[j].pos.x;
      dy = (float)vectors[i].pos.y - (float)vectors[j].pos.y;
      distance_1 = sqrt(dx * dx + dy * dy);

      // distance in current image:
      dx = (float)vectors[i].pos.x + (float)vectors[i].flow_x - (float)vectors[j].pos.x - (float)vectors[j].flow_x;
      dy = (float)vectors[i].pos.y + (float)vectors[i].flow_y - (float)vectors[j].pos.y - (float)vectors[j].flow_y;
      distance_2 = sqrt(dx * dx + dy * dy);


      // calculate divergence for this sample:
      divs[sample] = (distance_2 - distance_1) / distance_1;
    }

    // calculate the mean divergence:
    mean_divergence = get_mean(divs, n_samples);
    // free the memory of divs:
    free(divs);
  }

  // return the calculated divergence:
  return mean_divergence;
}

float get_mean(float *numbers, int n_elements)
{
  int i = 0;
  float mean = 0;
  for (i = 0; i < n_elements; i++) {
    mean += numbers[i];
  }
  mean /= n_elements;
  return mean;
}